ci(release): skip GitHub relay when GH_MIRROR_TOKEN absent

Relay is on hold until after the rc dress rehearsal — without the
secret the step exits cleanly instead of failing the release run.
Adding the secret later enables it with no workflow change.

.env.example

@@ -1,5 +1,5 @@
 # ─── uptop configuration ───────────────────────────────────
-# Export in your environment or pass via docker run --env-file.\n# Only uncomment what you need.
+# Copy to .env and edit. Only uncomment what you need.
 
 # ─── Core ──────────────────────────────────────────────────
 UPTOP_PORT=23234              # SSH server port
@@ -40,5 +40,3 @@ UPTOP_DB_DSN=/data/uptop.db   # File path (SQLite) or connection string (Postgre
 # UPTOP_ALLOW_PRIVATE_TARGETS=false    # Allow monitoring RFC1918/loopback addresses
 # UPTOP_METRICS_PUBLIC=false           # Expose /metrics without auth
 # UPTOP_CORS_ORIGIN=                   # Access-Control-Allow-Origin for /status/json
-# UPTOP_TRUSTED_PROXIES=              # Comma-separated CIDRs/IPs for X-Forwarded-For trust
-# UPTOP_MAINT_RETENTION=168h           # How long ended maintenance windows are kept

.gitea/issue_template/bug_report.yaml

@@ -16,7 +16,7 @@ body:
       label: What happened?
       description: Include what you expected to happen instead.
       placeholder: |
-        When I run `uptop`, the TUI crashes after 10 seconds.
+        When I run `uptop serve`, the TUI crashes after 10 seconds.
         I expected it to keep running and display monitor status.
     validations:
       required: true
@@ -25,7 +25,7 @@ body:
     attributes:
       label: Steps to reproduce
       placeholder: |
-        1. Run `uptop`
+        1. Run `uptop serve`
         2. Wait ~10 seconds
         3. TUI crashes with panic
     validations:
@@ -37,7 +37,7 @@ body:
       description: Output of `uptop version`, OS, terminal. Paste any errors below.
       render: shell
       placeholder: |
-        uptop 0.1.0 (abc1234, 2026-06-17)
+        uptop version 2026.06.1
         OS: Debian 13
         Terminal: Ghostty
 

.gitea/workflows/release-binaries.yml

@@ -49,12 +49,41 @@ jobs:
           version: "~> v2"
           args: release --clean --release-notes=/tmp/release-notes.md
         env:
-          GORELEASER_CURRENT_TAG: ${{ github.ref_name }}
           GORELEASER_FORCE_TOKEN: gitea
           GITEA_TOKEN: ${{ secrets.RELEASE_TOKEN }}
           GITEA_API_URL: http://gitea:3000/api/v1
-          HOMEBREW_TAP_GITHUB_TOKEN: ${{ secrets.HOMEBREW_TAP_GITHUB_TOKEN }}
 
-      # GitHub release relaying is handled by .github/workflows/mirror-release.yml,
-      # which runs on GitHub Actions when the push mirror delivers the tag and
-      # copies this run's Gitea release assets — no PAT needed on this side.
+      # GoReleaser publishes to exactly one SCM (Gitea). The push mirror
+      # carries git refs but not release artifacts, so relay the release to
+      # the GitHub mirror — README install links point there.
+      - name: Mirror release to GitHub
+        env:
+          GH_TOKEN: ${{ secrets.GH_MIRROR_TOKEN }}
+        run: |
+          if [ -z "$GH_TOKEN" ]; then
+            echo "GH_MIRROR_TOKEN not set — skipping GitHub release relay"
+            exit 0
+          fi
+          apk add --no-cache github-cli
+          TAG="${{ github.ref_name }}"
+
+          # Nudge the push mirror, then wait for the tag to land on GitHub.
+          curl -sf -X POST \
+            -H "Authorization: token ${{ secrets.RELEASE_TOKEN }}" \
+            "http://gitea:3000/api/v1/repos/lerkolabs/uptop/push_mirrors-sync" || true
+          for i in $(seq 1 30); do
+            if gh api "repos/lerkolabs/uptop/git/ref/tags/${TAG}" >/dev/null 2>&1; then
+              break
+            fi
+            sleep 10
+          done
+
+          PRERELEASE=""
+          case "$TAG" in *-*) PRERELEASE="--prerelease";; esac
+          gh release create "$TAG" \
+            --repo lerkolabs/uptop \
+            --verify-tag \
+            --title "$TAG" \
+            --notes-file /tmp/release-notes.md \
+            $PRERELEASE \
+            dist/*.tar.gz dist/*.zip dist/*.deb dist/*.rpm dist/checksums.txt

.gitea/workflows/release-docker.yml

@@ -35,17 +35,8 @@ jobs:
 
           TAGS="lerkolabs/uptop:${TAG}"
           TAGS="${TAGS},lerkolabs/uptop:sha-${SHORT_SHA}"
-          TAGS="${TAGS},ghcr.io/lerkolabs/uptop:${TAG}"
-          TAGS="${TAGS},ghcr.io/lerkolabs/uptop:sha-${SHORT_SHA}"
-          # :latest only for real releases — rc rehearsal tags must not move it
           if [ "${{ github.ref_type }}" = "tag" ]; then
-            case "$TAG" in
-              *-*) ;;
-              *)
-                TAGS="${TAGS},lerkolabs/uptop:latest"
-                TAGS="${TAGS},ghcr.io/lerkolabs/uptop:latest"
-                ;;
-            esac
+            TAGS="${TAGS},lerkolabs/uptop:latest"
           fi
           echo "tags=$TAGS" >> "$GITHUB_OUTPUT"
 
@@ -61,33 +52,6 @@ jobs:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
 
-      - name: Log in to GHCR
-        uses: docker/login-action@v3
-        with:
-          registry: ghcr.io
-          username: ${{ secrets.GHCR_USERNAME }}
-          password: ${{ secrets.GHCR_TOKEN }}
-
-      # Scan must gate the push: build amd64 locally, scan it, and only then run
-      # the multi-arch push (amd64 layers come from the builder cache, so the
-      # second build only adds the arm64 work).
-      - name: Build for scan (amd64, local)
-        uses: docker/build-push-action@v5
-        with:
-          context: .
-          load: true
-          platforms: linux/amd64
-          tags: uptop-scan:${{ steps.meta.outputs.tag }}
-          build-args: |
-            VERSION=${{ steps.meta.outputs.tag }}
-            COMMIT=${{ github.sha }}
-            BUILD_DATE=${{ github.event.head_commit.timestamp }}
-
-      - name: Scan image for CVEs
-        run: |
-          curl -sSfL https://raw.githubusercontent.com/anchore/grype/main/install.sh | sh -s -- -b /usr/local/bin v0.114.0
-          grype uptop-scan:${{ steps.meta.outputs.tag }} --fail-on critical --output table
-
       - name: Build and push
         uses: docker/build-push-action@v5
         with:
@@ -102,6 +66,11 @@ jobs:
             COMMIT=${{ github.sha }}
             BUILD_DATE=${{ github.event.head_commit.timestamp }}
 
+      - name: Scan image for CVEs
+        run: |
+          curl -sSfL https://raw.githubusercontent.com/anchore/grype/main/install.sh | sh -s -- -b /usr/local/bin v0.114.0
+          grype lerkolabs/uptop:${{ steps.meta.outputs.tag }} --fail-on critical --output table
+
       - name: Update Docker Hub description
         uses: peter-evans/dockerhub-description@v4
         with:
@@ -112,7 +81,5 @@ jobs:
       - name: Cleanup Docker artifacts
         if: always()
         run: |
-          # the scan image is tagged, so image prune won't catch it
-          docker image rm "uptop-scan:${{ steps.meta.outputs.tag }}" 2>/dev/null || true
           docker image prune -f
           docker builder prune -f --keep-storage=2GB

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -1,46 +0,0 @@
-name: Bug Report
-description: Something isn't working as expected
-labels:
-  - bug
-body:
-  - type: checkboxes
-    id: search
-    attributes:
-      label: Before filing
-      options:
-        - label: I searched existing issues and didn't find a match
-          required: true
-  - type: textarea
-    id: description
-    attributes:
-      label: What happened?
-      description: Include what you expected to happen instead.
-      placeholder: |
-        When I run `uptop`, the TUI crashes after 10 seconds.
-        I expected it to keep running and display monitor status.
-    validations:
-      required: true
-  - type: textarea
-    id: reproduction
-    attributes:
-      label: Steps to reproduce
-      placeholder: |
-        1. Run `uptop`
-        2. Wait ~10 seconds
-        3. TUI crashes with panic
-    validations:
-      required: true
-  - type: textarea
-    id: environment
-    attributes:
-      label: Environment & logs
-      description: Output of `uptop version`, OS, terminal. Paste any errors below.
-      render: shell
-      placeholder: |
-        uptop 0.1.0 (abc1234, 2026-06-17)
-        OS: Debian 13
-        Terminal: Ghostty
-
-        [paste any error output here]
-    validations:
-      required: false

.github/ISSUE_TEMPLATE/config.yml

@@ -1 +1,8 @@
 blank_issues_enabled: false
+contact_links:
+  - name: Report a Bug
+    url: https://gitea.lerkolabs.com/lerkolabs/uptop/issues/new?template=bug_report.yaml
+    about: Report bugs on our Gitea instance
+  - name: Request a Feature
+    url: https://gitea.lerkolabs.com/lerkolabs/uptop/issues/new?template=feature_request.yaml
+    about: Suggest features on our Gitea instance

.github/ISSUE_TEMPLATE/feature_request.yml

@@ -1,20 +0,0 @@
-name: Feature Request
-description: Suggest a new feature or enhancement
-labels:
-  - feature
-body:
-  - type: textarea
-    id: problem
-    attributes:
-      label: Problem
-      description: What's frustrating or missing?
-      placeholder: I find myself always needing to ...
-    validations:
-      required: true
-  - type: textarea
-    id: solution
-    attributes:
-      label: Proposed solution
-      description: How would you like this to work?
-    validations:
-      required: false

.github/workflows/forward-issues.yml

@@ -1,78 +0,0 @@
-name: Forward Issues to Gitea
-
-on:
-  issues:
-    types: [opened]
-
-permissions:
-  issues: write
-
-jobs:
-  forward:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Build issue body
-        env:
-          ISSUE_BODY: ${{ github.event.issue.body }}
-          ISSUE_URL: ${{ github.event.issue.html_url }}
-          ISSUE_AUTHOR: ${{ github.event.issue.user.login }}
-        run: |
-          jq -n \
-            --arg url "$ISSUE_URL" \
-            --arg author "$ISSUE_AUTHOR" \
-            --arg body "$ISSUE_BODY" \
-            '">" + " Forwarded from GitHub: " + $url + "\n> Reported by: [@" + $author + "](https://github.com/" + $author + ")\n\n---\n\n" + $body' \
-            -r > /tmp/gitea-issue-body.md || exit 1
-
-      - name: Resolve label IDs
-        id: labels
-        env:
-          GITEA_TOKEN: ${{ secrets.GITEA_TOKEN }}
-          GH_LABELS: ${{ toJSON(github.event.issue.labels.*.name) }}
-        run: |
-          GITEA_LABELS=$(curl -f \
-            -H "Authorization: token ${GITEA_TOKEN}" \
-            "https://gitea.lerkolabs.com/api/v1/repos/lerkolabs/uptop/labels?limit=50") || exit 1
-
-          GH_LABELS_COMPACT=$(echo "$GH_LABELS" | jq -c '.')
-          LABEL_IDS=$(echo "$GITEA_LABELS" | jq -c --argjson gh "$GH_LABELS_COMPACT" '
-            [.[] | select(
-              .name == "github" or
-              (.name as $n | $gh | index($n) != null)
-            ) | .id]
-          ')
-
-          echo "ids=${LABEL_IDS}" >> "$GITHUB_OUTPUT"
-
-      - name: Create Gitea issue
-        id: create
-        env:
-          GITEA_TOKEN: ${{ secrets.GITEA_TOKEN }}
-          ISSUE_TITLE: ${{ github.event.issue.title }}
-          LABEL_IDS: ${{ steps.labels.outputs.ids }}
-        run: |
-          RESPONSE=$(curl -f -X POST \
-            -H "Authorization: token ${GITEA_TOKEN}" \
-            -H "Content-Type: application/json" \
-            "https://gitea.lerkolabs.com/api/v1/repos/lerkolabs/uptop/issues" \
-            -d "$(jq -n \
-              --arg title "$ISSUE_TITLE" \
-              --rawfile body /tmp/gitea-issue-body.md \
-              --argjson labels "$LABEL_IDS" \
-              '{title: $title, body: $body, labels: $labels}'
-            )") || exit 1
-
-          GITEA_URL=$(echo "$RESPONSE" | jq -re '.html_url') || exit 1
-          GITEA_NUM=$(echo "$RESPONSE" | jq -re '.number') || exit 1
-
-          echo "url=${GITEA_URL}" >> "$GITHUB_OUTPUT"
-          echo "number=${GITEA_NUM}" >> "$GITHUB_OUTPUT"
-
-      - name: Comment on GitHub issue
-        env:
-          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          GITEA_URL: ${{ steps.create.outputs.url }}
-        run: |
-          gh issue comment "${{ github.event.issue.number }}" \
-            --repo "${{ github.repository }}" \
-            --body "Thanks for reporting! This issue has been forwarded to our [primary tracker](${GITEA_URL}). Discussion and updates will happen there."

.github/workflows/mirror-release.yml

@@ -19,35 +19,26 @@ jobs:
         run: |
           API="https://gitea.lerkolabs.com/api/v1/repos/lerkolabs/uptop/releases/tags/${TAG}"
 
-          # 40 x 30s = 20 min: the Gitea release can queue behind the ~18-min
-          # Docker job on the single runner. Asset count must hold steady for
-          # two consecutive polls — GoReleaser uploads one file at a time, and
-          # mirroring mid-upload would publish a partial asset set.
-          PREV_COUNT=0
-          ASSET_COUNT=0
-          for i in $(seq 1 40); do
+          for i in $(seq 1 20); do
             if RESPONSE=$(curl -sf "$API" 2>/dev/null); then
               ASSET_COUNT=$(echo "$RESPONSE" | jq '.assets | length')
-              if [ "$ASSET_COUNT" -gt 0 ] && [ "$ASSET_COUNT" -eq "$PREV_COUNT" ]; then
-                echo "Found release with $ASSET_COUNT assets (stable)"
+              if [ "$ASSET_COUNT" -gt 0 ]; then
+                echo "Found release with $ASSET_COUNT assets"
                 break
               fi
-              echo "Release has $ASSET_COUNT assets (was $PREV_COUNT)... attempt $i/40"
-              PREV_COUNT="$ASSET_COUNT"
+              echo "Release exists but no assets yet... attempt $i/20"
             else
-              echo "Waiting for Gitea release... attempt $i/40"
+              echo "Waiting for Gitea release... attempt $i/20"
             fi
             sleep 30
           done
 
           if [ -z "$RESPONSE" ] || [ "$ASSET_COUNT" -eq 0 ]; then
-            echo "::error::Gitea release for ${TAG} not found or has no assets after 20 minutes"
+            echo "::error::Gitea release for ${TAG} not found or has no assets after 10 minutes"
             exit 1
           fi
 
-          # select() so an empty-string body produces an empty file — `// empty`
-          # treats "" as truthy and wrote a blank line, defeating this fallback.
-          echo "$RESPONSE" | jq -r '.body | select(. != null and . != "")' > /tmp/release-notes.md
+          echo "$RESPONSE" | jq -r '.body // empty' > /tmp/release-notes.md
 
           if [ ! -s /tmp/release-notes.md ]; then
             echo "Release ${TAG} from [Gitea](https://gitea.lerkolabs.com/lerkolabs/uptop/releases/tag/${TAG})" > /tmp/release-notes.md
@@ -71,11 +62,8 @@ jobs:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           TAG: ${{ github.ref_name }}
         run: |
-          PRERELEASE=""
-          case "$TAG" in *-*) PRERELEASE="--prerelease" ;; esac
           gh release create "$TAG" \
             --repo "$GITHUB_REPOSITORY" \
             --title "$TAG" \
             --notes-file /tmp/release-notes.md \
-            $PRERELEASE \
             /tmp/assets/*

.gitignore

@@ -25,6 +25,4 @@ authorized_keys
 tmp
 *.local.json
 *.local.md
-data/
 .env
-vhs

.goreleaser.yaml

@@ -8,7 +8,6 @@ release:
   gitea:
     owner: lerkolabs
     name: uptop
-  prerelease: auto
 
 builds:
   - main: ./cmd/uptop
@@ -59,25 +58,5 @@ nfpms:
         dst: /usr/share/doc/uptop/LICENSE
         type: doc
 
-brews:
-  - name: uptop
-    repository:
-      owner: lerkolabs
-      name: homebrew-tap
-      token: "{{ .Env.HOMEBREW_TAP_GITHUB_TOKEN }}"
-    url_template: "https://github.com/lerkolabs/uptop/releases/download/{{ .Tag }}/{{ .ArtifactName }}"
-    commit_author:
-      name: lerkolabs-bot
-      email: bot@lerkolabs.com
-    homepage: https://github.com/lerkolabs/uptop
-    description: Self-hosted uptime monitoring with a TUI over SSH
-    license: MIT
-    install: |
-      bin.install "uptop"
-    test: |
-      system bin/"uptop", "version"
-
-# Changelog generation must stay enabled: the --release-notes flag is consumed
-# by the changelog pipe, so disabling it silently drops the git-cliff notes
-# (empty release body on v0.1.0-rc.1). With --release-notes set, GoReleaser
-# skips its own generation and uses the file.
+changelog:
+  disable: true

.grype.yaml

@@ -1,11 +1,6 @@
 ignore:
-  # SCP path traversal in charmbracelet/wish — same flaw, two ids: grype has
-  # matched it as CVE-2026-41589 and as GHSA-xjvp-7243-rg9h depending on db
-  # version, and ignore matching is exact-id, so both stay listed.
+  # CVE-2026-41589: SCP path traversal in charmbracelet/wish.
   # We only import wish/bubbletea for the SSH TUI server — the vulnerable
-  # scp.Middleware / scp.NewFileSystemHandler symbols are never compiled in
-  # (govulncheck reachability agrees). No fix for wish v1; v2
-  # (charm.land/wish/v2 >= 2.0.1) requires the bubbletea-v2 stack migration,
-  # tracked in issue #126. Remove both entries when that lands.
+  # scp.Middleware / scp.NewFileSystemHandler symbols are never compiled in.
+  # No fix available for wish v1; v2 (charm.land/wish/v2) patched in 2.0.1.
   - vulnerability: CVE-2026-41589
-  - vulnerability: GHSA-xjvp-7243-rg9h

CHANGELOG.md

@@ -1,183 +1,129 @@
 # Changelog
 
-## [Unreleased]
-
-### Added
-
-- initial commit — uptime monitor (forked from go-upkeep)
-- enhanced dashboard with lipgloss tables, huh forms, mouse support, and animations
-- upgrade users tab with lipgloss table, edit support, role select
-- upgrade alerts tab with lipgloss table, click zones, colored types
-- widen Site struct and DB schema for ping, port, dns, group monitor types
-- add ping, port, and DNS check routines
-- add ntfy notification provider with TUI support
-- add Uptime Kuma backup converter with CLI and API
-- add mouse wheel scrolling for all tabs
-- add per-site pause, fix viewport, polish status page
-- add monitor groups with collapse/expand and tree view
-- add Telegram, PagerDuty, Pushover, Gotify providers
-- add Prometheus /metrics endpoint
-- expose HTTP method and accepted status codes in monitor form
-- add config-as-code YAML import/export
-- add distributed probing foundation — schema, models, and probe APIs
-- add probe execution mode, check extraction, and result aggregation
-- add region affinity, Nodes TUI tab, and probe metrics
-- add status bar, tab badges, and detail panel
-- bordered modals, welcome state, and dynamic name width
-- DOWN-first sort, health pulse, and site filter
-- split available width evenly between NAME and HISTORY columns
-- add type icons to sites table
-- persist logs to DB, load on startup
-- add incident management and maintenance windows
-- zebra striping, detail breadcrumb, sparkline stats, collapse persistence
-- add --version flag with build metadata injection
-- add theme system with 4 curated palettes
-- swap light theme for Tokyo Night and Gruvbox
-- seed SSH users from env var and authorized_keys file (#31)
-- show error reason when monitors go DOWN
-- proper push monitor lifecycle — PENDING, LATE, DOWN states
-- logs tab overhaul — severity tags, filtering, recovery durations
-- alert channel health indicator + test alerts
-- add GitHub release relay workflow
-- classify error reasons on DOWN monitors
-- add state change history view with outage duration
-- add Opsgenie provider
-- add STALE state for push monitors
-- add SLA reporting view
-- overhaul latency sparkline scaling, color, and layout
-- auto-prune expired maintenance windows
-- click-to-inspect sparkline tooltips in detail view
+## [2026.06.2] — 2026-06-02 (infrastructure)
 
 ### Changed
-
-- replace database ID column with row counter
-- unify SQLite and Postgres into dialect-based SQLStore
-- add error returns to all Store interface methods
-- remove store global singleton, thread store explicitly
-- extract shared HTTPProvider for webhook-based alerts
-- extract shared table rendering, fix cursor bounds
-- encapsulate engine state, add graceful shutdown and tests
-- split release pipeline, add nfpm/homebrew/git-cliff
-- decompose god files into single-concern modules
-- consistent chrome across all views
-- status icons, clean STATUS column, relative time
-- extract magic numbers into named constants
-- check all discarded errors in sqlstore_test.go
-- overhaul tab bar — consistent counts, active highlight, colored alerts
-- responsive column hiding — 3-tier priority-based layout
-- swap mattn/go-sqlite3 for modernc.org/sqlite
-- propagate context.Context through all Store methods
-- typed Status constants with IsBroken() predicate
-- schema_version migration table + DeleteAlert FK fix
-- shared storetest.BaseMock replaces 5 duplicated mocks
-- consolidate env parsing into appConfig struct
-- extract Server type with named handler methods
-- split Site into SiteConfig + SiteState
-- unify logging with log/slog
-- restructure site form to 2 type-aware pages
+- Split release pipeline into separate binary and Docker workflows (#45)
+- Pin Docker base images by digest (#45)
+- Add GitHub release relay — mirrors Gitea releases to GitHub (#49)
+- Add Grype CVE scanning to Docker pipeline (#45)
+- Make CVE scan non-blocking for non-exploitable wish SCP vulnerability (#48)
 
 ### Fixed
+- git-cliff install in CI — resolve download URL dynamically, extract to /tmp (#46, #47)
 
-- forward all msg types to huh forms, improve row selection UX
-- harden TLS, timeouts, validation, logging, and token generation
-- add delete confirm, input validation, XSS fix, history persistence
-- correctness and robustness fixes across all subsystems
-- make status bar and tab badges visible
-- use stable sort to prevent site list shuffling each tick
-- sort children by ID before status to prevent map-order shuffling
-- sparkline now spans full column width
-- sparkline right-aligned — current time at right edge, dots fill left
-- increase history buffer to 60 so sparkline fills completely
-- compute uptime from windowed statuses, not running counters
-- seed status and latency from DB history on startup
-- strip push tokens from /status/json response
-- correct viewport sizing and dynamic chrome calculation
-- constrain form height to terminal and forward resize events
-- skip children in maintenance when computing group status
-- exclude maintenance'd monitors from down count and pulse
-- group selection highlight, layout constants, group history graphs
-- stable monitor count and universal group icons
-- replace panic with error return, handle unmarshal errors
-- add context to Provider.Send, log alert failures
-- constant-time secret comparison, request size limits
-- graceful shutdown for HTTP, SSH servers and database
-- add jitter to check intervals and stagger startup
-- use sh instead of bash for runner compatibility
-- enable CGO for race detector, use lint-action v7
-- install gcc for race detector support
-- skip irrelevant field validation by monitor type
-- guard max retries validator for group type
-- tighten zebra row contrast for Tokyo Night and Gruvbox
-- phase 1 critical fixes for public release
-- phase 2 high-severity hardening
-- phase 3 medium reliability and hardening
-- phase 4 code quality and low-severity fixes
-- rename GITEA_TOKEN to RELEASE_TOKEN
-- remove explicit container, use sh shell
-- bump golang.org/x/crypto v0.47.0 → v0.52.0
-- install git and gcc for GoReleaser in release pipeline
-- use internal Gitea URL for GoReleaser API calls
-- use docker-builder runner for Docker image builds
-- patch Docker Scout CVEs and remove unused openssh-client (#41)
-- non-root user, supply chain attestations, build cleanup
-- move SSH host key path into /data for non-root user
-- create .ssh dir explicitly, ensure entrypoint is executable
-- resolve git-cliff download URL dynamically
-- extract git-cliff to /tmp to avoid dirty worktree
-- make Grype CVE scan non-blocking for known wish vuln
-- bump Go 1.26.3 → 1.26.4
-- remove error truncation from detail panel
-- classify safedial "failed to connect" as TCP
-- resolve staticcheck lint errors in history view
-- trigger immediate recheck after site config edit
-- broken tick chain after form/dialog + retries off-by-one
-- wire up [e] edit key in detail panel
-- show push token and URL in detail panel
-- show correct push heartbeat curl command in detail panel
-- propagate STALE/LATE child status to group
-- quick wins batch — version footer, column widths, zebra, sparkline
-- logs tab use viewport for scrollable content
-- pin footer to bottom of terminal
-- normalize content whitespace for consistent footer position
-- clip overflowing content to keep footer pinned
-- remove extra blank lines above footer
-- expand log viewport to fill content area
-- log STALE recovery in push heartbeat handler
-- check fmt.Sscanf return value (errcheck lint)
-- inject time into ComputeDailyBreakdown for testability
-- cascade delete related rows when removing a site
-- merge check results into live state, never overwrite
-- serialize DB writes through a single drained writer
-- close XFF bypass and three secret-leak paths
-- move blocking DB IO out of Update/View into tea.Cmds
-- move theme styles onto the Model to end cross-session races
-- finish moving keypress DB reads into tea.Cmds
-- move all store writes out of Update into tea.Cmds
-- mask alert secrets in the TUI detail panel and table
-- serve /status/json through a public DTO
-- make SSH key revocation fail closed
-- six correctness fixes for the state machine
-- migrate Postgres timestamps to TIMESTAMPTZ
-- seven quick-win bug fixes across engine, server, TUI, CLI
-- SSRF guard gaps + DNS port restriction + metrics auth
-- track selection by site ID + q means back everywhere
-- apply convergence + push/group check history
-- Kuma import tokens/paused, Docker hardening, migrate-secrets idempotency
-- six small fixes — rate limiter leak, DST SLA, probe sort, TUI cleanup
-- seven fixes — token scan, variadic cleanup, TUI layout, compose secrets
-- chmod SQLite DB files to 0600 on open
-- close DNS-rebind TOCTOU on ping/port checks
-- API import no longer replaces user accounts
-- email send respects context deadline
-- rename X-Upkeep-Secret header to X-Uptop-Secret
-- apply log filter to full log list, not viewport window
-- repair pipeline defects found in v0.1.0-rc.1 rehearsal
-- suppress wish GHSA alias in grype, fold rc tags into launch notes
-- scan gates docker push, rc tags spare :latest, mirror waits for stable assets
-- remove tagged scan image in cleanup step
-- exclude rc tags from cliff tag_pattern so launch notes span full history
-- fall back to embedded build info when ldflags absent
-- drop body-grep Security grouping, map polish type in cliff
-- sync selectedID on click so refreshLive doesn't revert cursor
-- resolve 4 tag-blocking issues for v0.1.0
+## [2026.06.1] — 2026-06-01
 
+### Changed
+- Container runs as non-root user `uptop` (UID/GID 1000) instead of root (#44)
+- SSH host key relocated to `/data/.ssh/id_ed25519` for non-root compatibility (#44)
+- Release workflow prunes dangling images and build cache after Docker push (#44)
+
+### Added
+- SBOM and provenance attestations on Docker images for supply chain compliance (#44)
+- Entrypoint script with volume writability check and migration guidance (#44)
+
+### Breaking
+- Existing Docker volumes with root-owned files require migration before upgrading:
+  `docker run --rm -v <volume>:/data alpine chown -R 1000:1000 /data`
+
+## [2026.05.6] — 2026-05-30 (infrastructure)
+
+### Changed
+- Sync README to Docker Hub on release (#43)
+
+### Security
+- Patch Docker Scout CVEs, remove unused openssh-client (#41)
+
+## [2026.05.5] — 2026-05-29
+
+### Added
+- Error reason display when monitors go DOWN (#33)
+- Push monitor lifecycle — PENDING, LATE, DOWN states (#34)
+- Logs tab overhaul — severity tags, filtering, recovery durations (#35)
+- Alert channel health indicator and test alerts (#36)
+- TUI screenshots in `assets/` (#32)
+- CI status badge in README
+
+### Changed
+- Visual polish — detail sections, column headers, alert detail (#37)
+- README rewritten with hero image, badges, collapsible install sections (#32)
+- Changelog rewritten to match actual CalVer tag history
+- Migrated to `lerkolabs` org namespace (#38)
+- Docker-compose files moved to `deploy/`
+
+## [2026.05.4] — 2026-05-27
+
+### Added
+- SSH user seeding from `UPTOP_ADMIN_KEY` env var and `UPTOP_KEYS` file (#31)
+- GoReleaser for binary releases
+- govulncheck in CI pipeline
+- Multi-arch Docker builds (amd64 + arm64)
+
+### Changed
+- CI overhaul — Go 1.26, build caching, streamlined pipeline (#30)
+- Bumped golang.org/x/crypto v0.47.0 → v0.52.0
+- Bumped Alpine 3.21 → 3.23
+
+### Security
+- Phase 1: SSRF protection, input validation, safe dial (#26)
+- Phase 2: TLS hardening, auth bypass fixes, rate limiting (#27)
+- Phase 3: Graceful degradation, connection limits, timeout enforcement (#28)
+- Phase 4: Code quality, error handling, linter fixes (#29)
+
+## [2026.05.3] — 2026-05-25
+
+### Added
+- Theme system with 5 dark palettes — Default, Dracula, Nord, Tokyo Night, Gruvbox (#24)
+- `--version` flag with build metadata injection
+- Gitea Actions CI pipeline — test + lint (#20)
+- golangci-lint configuration
+- Comprehensive test suite — 94 tests across monitor, server, cluster (#19)
+- CONTRIBUTING.md and SECURITY.md
+
+### Changed
+- Renamed project from go-upkeep to uptop (#25)
+- Updated LICENSE with dual copyright for independent fork
+
+### Fixed
+- Form validators scoped to relevant monitor types (#23)
+- Graceful shutdown for HTTP, SSH servers and database (#19)
+- Constant-time secret comparison, request size limits (#19)
+- Check interval jitter to prevent thundering herd (#19)
+- TUI visual polish — zebra striping, group icons, sparkline stats (#18)
+
+## [2026.05.2] — 2026-05-22
+
+### Added
+- Incident management and maintenance windows (#17)
+- Production docker-compose.yml
+
+### Fixed
+- Viewport sizing and dynamic chrome calculation (#16)
+- Form height constrained to terminal with resize forwarding
+- Maintenance'd monitors excluded from down count and pulse
+- Group status correctly skips children in maintenance
+
+## [2026.05.1] — 2026-05-16
+
+### Added
+- Distributed probing with leader + probe nodes
+- Config-as-code — YAML apply/export with dry-run and prune
+- TUI polish — status bar, tab badges, detail panel, modals
+- DOWN-first sort, health pulse, site filter
+- Type icons in sites table
+- Sparkline history graphs
+- Persistent state — uptime, status, latency, and logs survive restarts
+- Push token stripping from /status/json response
+
+## [2026.04.1] — 2026-04-01
+
+### Added
+- SSH-accessible TUI built on Bubble Tea + Wish
+- 6 check types — HTTP, Push, Ping, Port, DNS, Group
+- 9 alert providers — Discord, Slack, Email, Ntfy, Telegram, PagerDuty, Pushover, Gotify, Webhook
+- SQLite and PostgreSQL support
+- HA clustering with automatic failover
+- Prometheus /metrics endpoint
+- Public status page (HTML + JSON)
+- Uptime Kuma backup import

CONTRIBUTING.md

@@ -3,7 +3,7 @@
 ## Development
 
 ```sh
-go run ./cmd/uptop -demo        # starts with sample data
+go run cmd/uptop/main.go -demo  # starts with sample data
 ssh -p 23234 localhost              # connect to TUI
 ```
 

Dockerfile

@@ -1,5 +1,5 @@
 # --- Stage 1: Builder ---
-FROM golang:1.26.5-alpine3.23@sha256:622e56dbc11a8cfe87cafa2331e9a201877271cbff918af53d3be315f3da88cc AS builder
+FROM golang:1.26-alpine3.23@sha256:91eda9776261207ea25fd06b5b7fed8d397dd2c0a283e77f2ab6e91bfa71079d AS builder
 WORKDIR /app
 COPY go.mod go.sum ./
 RUN --mount=type=cache,target=/go/pkg/mod \
@@ -23,13 +23,14 @@ RUN mkdir -p /data/.ssh && chown -R uptop:uptop /data
 COPY --from=builder /app/uptop .
 COPY --chmod=755 docker-entrypoint.sh /usr/local/bin/
 
+ENV LIPGLOSS_RENDERER_HAS_DARK_BACKGROUND=true
 ENV UPTOP_DB_TYPE=sqlite
 ENV UPTOP_DB_DSN=/data/uptop.db
 ENV UPTOP_KEYS=/data/authorized_keys
 ENV UPTOP_SSH_HOST_KEY=/data/.ssh/id_ed25519
 ENV UPTOP_PORT=23234
 
-EXPOSE 8080 23234
+EXPOSE 23234
 HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=3 \
   CMD wget -qO- http://localhost:8080/api/health || exit 1
 USER uptop

README.md

@@ -1,18 +1,16 @@
 <div align="center">
-  <img src="assets/logo.svg" alt="uptop" width="320">
+  <h1>uptop</h1>
   <p>Self-hosted uptime monitoring with a TUI over SSH.</p>
   <p>No browser. No client install. Just <code>ssh -p 23234 your-server</code>.</p>
 
   <p>
-    <a href="https://gitea.lerkolabs.com/lerkolabs/uptop/actions"><img src="https://gitea.lerkolabs.com/lerkolabs/uptop/actions/workflows/ci.yml/badge.svg" alt="CI"></a>
-    <a href="https://github.com/lerkolabs/uptop/releases/latest"><img src="https://img.shields.io/github/v/release/lerkolabs/uptop" alt="Latest Release"></a>
-    <a href="https://goreportcard.com/report/github.com/lerkolabs/uptop"><img src="https://goreportcard.com/badge/github.com/lerkolabs/uptop" alt="Go Report Card"></a>
+    <a href="https://gitea.lerkolabs.com/lerkolabs/uptop/actions/workflows/ci.yml"><img src="https://gitea.lerkolabs.com/lerkolabs/uptop/actions/workflows/ci.yml/badge.svg" alt="CI"></a>
+    <img src="https://img.shields.io/badge/license-MIT-blue" alt="MIT License">
     <img src="https://img.shields.io/badge/go-1.26-00ADD8?logo=go&logoColor=white" alt="Go 1.26">
     <img src="https://img.shields.io/docker/pulls/lerkolabs/uptop" alt="Docker Pulls">
-    <img src="https://img.shields.io/badge/license-MIT-blue" alt="MIT License">
   </p>
 
-  <img src="assets/demo.gif" alt="uptop demo" width="800">
+  <img src="assets/monitors.png" alt="uptop monitors view" width="800">
 </div>
 
 ## What is this
@@ -21,57 +19,44 @@ An uptime monitor you manage entirely from the terminal. It runs as a server, ex
 
 Built on [RDGames/go-upkeep](https://github.com/RDGames/go-upkeep). Rewritten for clustering, config-as-code, and a proper dashboard.
 
-Canonical repo: [gitea.lerkolabs.com/lerkolabs/uptop](https://gitea.lerkolabs.com/lerkolabs/uptop) — [GitHub](https://github.com/lerkolabs/uptop) is a mirror; releases are published to both.
-
 ## Features
 
 - **6 check types** — HTTP, Push (heartbeat), Ping, Port, DNS, Groups
 - **10 alert providers** — Discord, Slack, Email, Ntfy, Webhook, Telegram, PagerDuty, Pushover, Gotify, Opsgenie
 - **Config as code** — define monitors in YAML, apply declaratively, version control your setup
 - **HA clustering** — leader/follower with automatic failover
-- **Prometheus metrics** — `/metrics` endpoint (`UPTOP_METRICS_PUBLIC=true` to expose without auth)
+- **Prometheus metrics** — `/metrics` endpoint, wire it straight to Grafana
 - **Public status page** — HTML + JSON, toggle with an env var
 - **SQLite or Postgres** — SQLite for single-node, Postgres for production
 - **Uptime Kuma import** — migrate from Kuma with one command
 
-> Group monitors roll up child status for display but don't fire their own alerts yet — attach alerts to the children.
-
 ## Screenshots
 
 <table>
   <tr>
-    <td><img src="assets/monitors.png" alt="monitors + log sidebar" width="400"></td>
-    <td><img src="assets/detail.png" alt="inline detail panel" width="400"></td>
+    <td><img src="assets/detail.png" alt="detail panel" width="400"></td>
+    <td><img src="assets/alerts.png" alt="alerts view" width="400"></td>
   </tr>
   <tr>
-    <td><img src="assets/fulldetail.png" alt="full-screen detail" width="400"></td>
-    <td><img src="assets/settings.png" alt="settings / alerts" width="400"></td>
-  </tr>
-  <tr>
-    <td><img src="assets/maint.png" alt="maintenance windows" width="400"></td>
+    <td><img src="assets/logs.png" alt="logs view" width="400"></td>
     <td><img src="assets/nodes.png" alt="cluster nodes" width="400"></td>
   </tr>
+  <tr>
+    <td colspan="2" align="center"><img src="assets/theme.png" alt="theme selection" width="600"></td>
+  </tr>
 </table>
 
-### Themes
-
-Five built-in themes: Flexoki Dark, Tokyo Night, Catppuccin Mocha, Nord, Gruvbox. Press `T` to cycle.
-
-<p align="center">
-  <img src="assets/themes.png" alt="all five themes" width="800">
-</p>
-
 ## Quick start
 
 ```bash
-UPTOP_ADMIN_KEY="$(cat ~/.ssh/id_ed25519.pub)" go run ./cmd/uptop
+go run cmd/uptop/main.go
 ssh -p 23234 localhost
 ```
 
 Want some data to look at first:
 
 ```bash
-UPTOP_ADMIN_KEY="$(cat ~/.ssh/id_ed25519.pub)" go run ./cmd/uptop -demo
+go run cmd/uptop/main.go -demo
 ```
 
 ## Install
@@ -94,20 +79,16 @@ services:
       # - UPTOP_ADMIN_KEY=ssh-ed25519 AAAA... you@host
     volumes:
       - ./data:/data
-    sysctls:
-      - net.ipv4.ping_group_range=0 2147483647
 ```
 
-First run: set `UPTOP_ADMIN_KEY` to your SSH public key.
-
-The `sysctls` line enables unprivileged ICMP inside the container — without it, ping monitors get no response and silently report DOWN.
+First run: set `UPTOP_ADMIN_KEY` to your SSH public key, or attach to the container and add it in the Users tab.
 
 </details>
 
 <details>
-<summary><strong>Binary (Linux, macOS, Windows)</strong></summary>
+<summary><strong>Binary (Linux amd64)</strong></summary>
 
-Download from [Releases](https://github.com/lerkolabs/uptop/releases) — amd64 and arm64 tarballs (zip for Windows), plus `.deb`/`.rpm` packages and `checksums.txt`.
+Download from [Releases](https://github.com/lerkolabs/uptop/releases).
 
 </details>
 
@@ -157,8 +138,6 @@ Full reference in [docs/config-as-code.md](docs/config-as-code.md).
 | `UPTOP_INSECURE_SKIP_VERIFY` | `false` | Skip TLS verification for checks |
 | `UPTOP_ALLOW_PRIVATE_TARGETS` | `false` | Allow monitoring RFC1918/loopback addresses |
 | `UPTOP_ADMIN_KEY` | | SSH public key seeded as first admin on startup |
-| `UPTOP_METRICS_PUBLIC` | `false` | Expose `/metrics` without auth |
-| `UPTOP_MAINT_RETENTION` | `168h` | How long ended maintenance windows are kept |
 | `UPTOP_TRUSTED_PROXIES` | | Comma-separated CIDRs/IPs whose `X-Forwarded-For` is trusted ([details](#running-behind-a-reverse-proxy)) |
 
 See [`.env.example`](.env.example) for all options including TLS, probes, and advanced settings.
@@ -183,19 +162,6 @@ Set `UPTOP_ENCRYPTION_KEY` to encrypt alert credentials (SMTP passwords, webhook
 
 Without this, credentials are stored as plaintext in the database. uptop warns on startup if unset. To encrypt credentials on an existing install, run `uptop migrate-secrets` with the key set.
 
-### Data retention
-
-uptop prunes its own history in the background — no external cleanup jobs needed:
-
-| Data | Kept |
-|---|---|
-| Check history | newest 1,000 checks per monitor |
-| State changes (UP/DOWN transitions) | newest 5,000 per monitor |
-| Logs | newest 200 entries |
-| Maintenance windows | 7 days after they end (`UPTOP_MAINT_RETENTION`) |
-
-Sparklines, uptime percentages, and SLA reports are computed from these windows, so very long-horizon stats aren't retained. Export to Prometheus via `/metrics` if you need unlimited history.
-
 ## Clustering
 
 uptop supports three modes: **leader** (default single node), **follower** (HA failover — takes over if the leader goes down), and **probe** (stateless distributed checks from multiple regions).
@@ -208,7 +174,7 @@ Export your Kuma backup JSON, then:
 
 ```bash
 curl -X POST http://localhost:8080/api/import/kuma \
-  -H "X-Uptop-Secret: your-secret" \
+  -H "X-Upkeep-Secret: your-secret" \
   -H "Content-Type: application/json" \
   -d @kuma-backup.json
 ```

assets/logo.svg

@@ -1,10 +0,0 @@
-<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 320 80" fill="none">
-  <!-- Terminal-inspired text mark for uptop -->
-  <rect width="320" height="80" rx="8" fill="#1C1B1A"/>
-  <!-- Prompt caret -->
-  <text x="16" y="52" font-family="'JetBrains Mono','Fira Code','SF Mono',monospace" font-size="36" font-weight="700" fill="#3AA99F">▲</text>
-  <!-- "uptop" in monospace -->
-  <text x="52" y="52" font-family="'JetBrains Mono','Fira Code','SF Mono',monospace" font-size="36" font-weight="700" fill="#CECDC3">uptop</text>
-  <!-- Status dot -->
-  <circle cx="296" cy="40" r="6" fill="#879A39"/>
-</svg>

cliff.toml

@@ -23,13 +23,7 @@ filter_unconventional = true
 split_commits = false
 protect_breaking_commits = false
 filter_commits = false
-# Only final tags count as releases — rc rehearsal tags must not become
-# section boundaries, or the final tag's notes would cover only
-# commits-since-last-rc (v0.1.0 rendered 0 commits with ignore_tags, which
-# drops rc-tagged commits instead of folding them forward). With rc tags
-# outside the pattern, finals render the full span and rc tags render
-# [Unreleased] with everything pending. Verified empirically on both.
-tag_pattern = 'v[0-9]+\.[0-9]+\.[0-9]+$'
+tag_pattern = "v[0-9].*"
 topo_order = false
 sort_commits = "oldest"
 
@@ -39,7 +33,7 @@ commit_parsers = [
     { message = "^perf", group = "Changed" },
     { message = "^refactor", group = "Changed" },
     { message = "^security", group = "Security" },
-    { message = "^polish", group = "Changed" },
+    { body = ".*security", group = "Security" },
     { body = "BREAKING", group = "Breaking" },
     { footer = "BREAKING.CHANGE", group = "Breaking" },
     { message = "^docs", skip = true },

cmd/uptop/keycache_test.go

@@ -43,10 +43,10 @@ func TestKeyCache_AllowsKnownDeniesUnknown(t *testing.T) {
 	_, unknown := testKey(t)
 	kc := newKeyCache(&kcMockStore{users: []models.User{{PublicKey: authorized}}})
 
-	if !kc.IsAllowed(context.Background(), known) {
+	if !kc.IsAllowed(known) {
 		t.Error("known key denied")
 	}
-	if kc.IsAllowed(context.Background(), unknown) {
+	if kc.IsAllowed(unknown) {
 		t.Error("unknown key allowed")
 	}
 }
@@ -56,7 +56,7 @@ func TestKeyCache_RetainsKeysOnRefreshError(t *testing.T) {
 	ms := &kcMockStore{users: []models.User{{PublicKey: authorized}}}
 	kc := newKeyCache(ms)
 
-	if !kc.IsAllowed(context.Background(), known) {
+	if !kc.IsAllowed(known) {
 		t.Fatal("known key denied on first refresh")
 	}
 
@@ -67,7 +67,7 @@ func TestKeyCache_RetainsKeysOnRefreshError(t *testing.T) {
 	kc.updated = time.Now().Add(-time.Hour)
 	kc.mu.Unlock()
 
-	if !kc.IsAllowed(context.Background(), known) {
+	if !kc.IsAllowed(known) {
 		t.Error("transient refresh error locked out a previously valid key")
 	}
 }
@@ -77,7 +77,7 @@ func TestKeyCache_FailsClosedAfterInvalidate(t *testing.T) {
 	ms := &kcMockStore{users: []models.User{{PublicKey: authorized}}}
 	kc := newKeyCache(ms)
 
-	if !kc.IsAllowed(context.Background(), known) {
+	if !kc.IsAllowed(known) {
 		t.Fatal("known key denied on first refresh")
 	}
 
@@ -86,7 +86,7 @@ func TestKeyCache_FailsClosedAfterInvalidate(t *testing.T) {
 	ms.err = errors.New("db down")
 	kc.Invalidate()
 
-	if kc.IsAllowed(context.Background(), known) {
+	if kc.IsAllowed(known) {
 		t.Error("revoked key still allowed while DB is down — fails open")
 	}
 }
@@ -97,7 +97,7 @@ func TestUserInvalidatingStore_DeleteDropsKeyCache(t *testing.T) {
 	kc := newKeyCache(ms)
 	s := &userInvalidatingStore{Store: ms, kc: kc}
 
-	if !kc.IsAllowed(context.Background(), known) {
+	if !kc.IsAllowed(known) {
 		t.Fatal("known key denied on first refresh")
 	}
 
@@ -109,7 +109,7 @@ func TestUserInvalidatingStore_DeleteDropsKeyCache(t *testing.T) {
 	ms.users = nil
 	ms.err = errors.New("db down")
 
-	if kc.IsAllowed(context.Background(), known) {
+	if kc.IsAllowed(known) {
 		t.Error("deleted user's key still allowed from stale cache")
 	}
 }

cmd/uptop/main.go

@@ -12,7 +12,6 @@ import (
 	"os"
 	"os/signal"
 	"path/filepath"
-	"runtime/debug"
 	"strings"
 	"sync"
 	"syscall"
@@ -40,37 +39,6 @@ var (
 	date    = "unknown"
 )
 
-// GoReleaser stamps the vars above via ldflags, but `go install module@tag`
-// compiles without them and would report "dev". The module version and any
-// vcs stamps are embedded in every binary, so fall back to those.
-func init() {
-	if version != "dev" {
-		return
-	}
-	info, ok := debug.ReadBuildInfo()
-	if !ok {
-		return
-	}
-	if mv := info.Main.Version; mv != "" && mv != "(devel)" {
-		mv = strings.TrimPrefix(mv, "v")
-		// Pseudo-versions (e.g. "0.1.1-0.20260620165311-5ca534b0b100+dirty")
-		// are noisy in the TUI footer. Extract just the base semver.
-		if i := strings.Index(mv, "-0."); i > 0 {
-			mv = mv[:i]
-		}
-		mv = strings.TrimSuffix(mv, "+dirty")
-		version = mv
-	}
-	for _, s := range info.Settings {
-		switch s.Key {
-		case "vcs.revision":
-			commit = s.Value
-		case "vcs.time":
-			date = s.Value
-		}
-	}
-}
-
 func main() {
 	slog.SetDefault(slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{
 		Level: slog.LevelInfo,
@@ -84,50 +52,23 @@ func main() {
 		case "export":
 			runExport(os.Args[2:])
 			return
-		case "version", "--version", "-v", "-version":
+		case "version", "--version", "-v":
 			printVersion()
 			return
 		case "migrate-secrets":
 			runMigrateSecrets(os.Args[2:])
 			return
-		case "help", "--help", "-h":
-			printUsage()
-			return
-		case "serve":
-			runServe(os.Args[2:])
-			return
 		}
 	}
 	runServe(os.Args[1:])
 }
 
-func printUsage() {
-	fmt.Fprintf(os.Stderr, `Usage: uptop <command> [flags]
-
-Commands:
-  serve             Start the server (default if no command given)
-  apply             Apply monitors from a YAML file
-  export            Export monitors to YAML
-  migrate-secrets   Re-encrypt alert credentials with current key
-  version           Print version and exit
-
-Run 'uptop serve --help' for server flags.
-`)
-}
-
 func printVersion() {
-	out := "uptop " + version
-	var meta []string
-	if commit != "none" {
-		meta = append(meta, commit)
+	if version == "dev" {
+		fmt.Println("uptop dev")
+	} else {
+		fmt.Printf("uptop %s (%s, %s)\n", version, commit, date)
 	}
-	if date != "unknown" {
-		meta = append(meta, date)
-	}
-	if len(meta) > 0 {
-		out += " (" + strings.Join(meta, ", ") + ")"
-	}
-	fmt.Println(out)
 }
 
 func envOrDefault(key, fallback string) string {
@@ -179,7 +120,7 @@ func parseTrustedProxies(raw string) []*net.IPNet {
 	return cidrs
 }
 
-func openStore(ctx context.Context, dbType, dsn string) store.Store {
+func openStore(dbType, dsn string) store.Store {
 	var ss *store.SQLStore
 	var err error
 	if dbType == "postgres" {
@@ -201,7 +142,7 @@ func openStore(ctx context.Context, dbType, dsn string) store.Store {
 	} else {
 		slog.Warn("no UPTOP_ENCRYPTION_KEY set, alert credentials stored unencrypted")
 	}
-	if err := ss.Init(ctx); err != nil {
+	if err := ss.Init(context.Background()); err != nil {
 		slog.Error("database init failed", "err", err)
 		os.Exit(1)
 	}
@@ -213,7 +154,7 @@ func runApply(args []string) {
 	filePath := fs.String("f", "", "Path to YAML config file (required)")
 	dryRun := fs.Bool("dry-run", false, "Show planned changes without applying")
 	prune := fs.Bool("prune", false, "Delete monitors/alerts not in YAML")
-	dbType := fs.String("db-type", envOrDefault("UPTOP_DB_TYPE", "sqlite"), "Database type (sqlite or postgres)")
+	dbType := fs.String("db-type", envOrDefault("UPTOP_DB_TYPE", "sqlite"), "Database type")
 	dsn := fs.String("dsn", envOrDefault("UPTOP_DB_DSN", "uptop.db"), "Database DSN")
 	_ = fs.Parse(args) // ExitOnError: parse errors exit before returning
 
@@ -223,8 +164,7 @@ func runApply(args []string) {
 		os.Exit(1)
 	}
 
-	ctx := context.Background()
-	s := openStore(ctx, *dbType, *dsn)
+	s := openStore(*dbType, *dsn)
 
 	f, err := config.LoadFile(*filePath)
 	if err != nil {
@@ -232,7 +172,7 @@ func runApply(args []string) {
 		os.Exit(1)
 	}
 
-	changes, err := config.Apply(ctx, s, f, config.ApplyOpts{
+	changes, err := config.Apply(context.Background(), s, f, config.ApplyOpts{
 		DryRun: *dryRun,
 		Prune:  *prune,
 	})
@@ -247,14 +187,13 @@ func runApply(args []string) {
 func runExport(args []string) {
 	fs := flag.NewFlagSet("export", flag.ExitOnError)
 	outPath := fs.String("o", "-", "Output file path (- for stdout)")
-	dbType := fs.String("db-type", envOrDefault("UPTOP_DB_TYPE", "sqlite"), "Database type (sqlite or postgres)")
+	dbType := fs.String("db-type", envOrDefault("UPTOP_DB_TYPE", "sqlite"), "Database type")
 	dsn := fs.String("dsn", envOrDefault("UPTOP_DB_DSN", "uptop.db"), "Database DSN")
 	_ = fs.Parse(args) // ExitOnError: parse errors exit before returning
 
-	ctx := context.Background()
-	s := openStore(ctx, *dbType, *dsn)
+	s := openStore(*dbType, *dsn)
 
-	f, err := config.Export(ctx, s)
+	f, err := config.Export(context.Background(), s)
 	if err != nil {
 		slog.Error("export failed", "err", err)
 		os.Exit(1)
@@ -268,7 +207,7 @@ func runExport(args []string) {
 
 func runMigrateSecrets(args []string) {
 	fs := flag.NewFlagSet("migrate-secrets", flag.ExitOnError)
-	dbType := fs.String("db-type", envOrDefault("UPTOP_DB_TYPE", "sqlite"), "Database type (sqlite or postgres)")
+	dbType := fs.String("db-type", envOrDefault("UPTOP_DB_TYPE", "sqlite"), "Database type")
 	dsn := fs.String("dsn", envOrDefault("UPTOP_DB_DSN", "uptop.db"), "Database DSN")
 	_ = fs.Parse(args)
 
@@ -283,8 +222,6 @@ func runMigrateSecrets(args []string) {
 		os.Exit(1)
 	}
 
-	ctx := context.Background()
-
 	var ss *store.SQLStore
 	if *dbType == "postgres" {
 		ss, err = store.NewPostgresStore(*dsn)
@@ -295,21 +232,21 @@ func runMigrateSecrets(args []string) {
 		slog.Error("database connection failed", "err", err)
 		os.Exit(1)
 	}
-	if err := ss.Init(ctx); err != nil {
+	if err := ss.Init(context.Background()); err != nil {
 		slog.Error("database init failed", "err", err)
 		os.Exit(1)
 	}
 
 	ss.SetEncryptor(enc)
 
-	alerts, err := ss.GetAllAlerts(ctx)
+	alerts, err := ss.GetAllAlerts(context.Background())
 	if err != nil {
 		slog.Error("failed to load alerts", "err", err)
 		os.Exit(1)
 	}
 	migrated := 0
 	for _, a := range alerts {
-		if err := ss.UpdateAlert(ctx, a.ID, a.Name, a.Type, a.Settings); err != nil {
+		if err := ss.UpdateAlert(context.Background(), a.ID, a.Name, a.Type, a.Settings); err != nil {
 			slog.Error("alert migration failed", "alert", a.Name, "err", err)
 			os.Exit(1)
 		}
@@ -362,7 +299,7 @@ func runServe(args []string) {
 
 	fs := flag.NewFlagSet("serve", flag.ExitOnError)
 	port := fs.Int("port", cfg.Port, "SSH Port")
-	flagDBType := fs.String("db-type", cfg.DBType, "Database type (sqlite or postgres)")
+	flagDBType := fs.String("db-type", cfg.DBType, "Database type")
 	flagDSN := fs.String("dsn", cfg.DBDSN, "Database DSN")
 	demo := fs.Bool("demo", false, "Seed demo data")
 	importKuma := fs.String("import-kuma", "", "Import Uptime Kuma backup JSON file")
@@ -396,19 +333,15 @@ func runServe(args []string) {
 
 	kc := newKeyCache(ss)
 	var s store.Store = &userInvalidatingStore{Store: ss, kc: kc}
-
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-
-	if err := s.Init(ctx); err != nil {
+	if err := s.Init(context.Background()); err != nil {
 		slog.Error("database init failed", "err", err)
 		os.Exit(1)
 	}
 	if *demo {
-		seedDemoData(ctx, s)
+		seedDemoData(s)
 	}
 
-	seedKeysFromEnv(ctx, s)
+	seedKeysFromEnv(s)
 
 	if *importKuma != "" {
 		kb, err := importer.LoadKumaFile(*importKuma)
@@ -417,7 +350,7 @@ func runServe(args []string) {
 			os.Exit(1)
 		}
 		backup := importer.ConvertKuma(kb)
-		if err := s.ImportData(ctx, backup); err != nil {
+		if err := s.ImportData(context.Background(), backup); err != nil {
 			slog.Error("import failed", "err", err)
 			os.Exit(1)
 		}
@@ -437,9 +370,12 @@ func runServe(args []string) {
 	}
 	eng.SetMaintRetention(cfg.MaintRetention)
 
-	eng.InitHistory(ctx)
-	eng.InitLogs(ctx)
-	eng.InitAlertHealth(ctx)
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	eng.InitHistory()
+	eng.InitLogs()
+	eng.InitAlertHealth()
 	eng.Start(ctx)
 
 	localTUI := isatty.IsTerminal(os.Stdout.Fd()) || isatty.IsCygwinTerminal(os.Stdout.Fd())
@@ -455,7 +391,7 @@ func runServe(args []string) {
 	sshSrv := startSSHServer(*port, s, eng, kc)
 
 	if localTUI {
-		p := tea.NewProgram(tui.InitialModel(ctx, true, s, eng, version), tea.WithAltScreen(), tea.WithMouseCellMotion())
+		p := tea.NewProgram(tui.InitialModel(true, s, eng, version), tea.WithAltScreen(), tea.WithMouseCellMotion())
 		if _, err := p.Run(); err != nil {
 			slog.Error("TUI failed", "err", err)
 		}
@@ -489,11 +425,11 @@ func startSSHServer(port int, db store.Store, eng *monitor.Engine, kc *keyCache)
 		wish.WithAddress(fmt.Sprintf(":%d", port)),
 		wish.WithHostKeyPath(envOrDefault("UPTOP_SSH_HOST_KEY", ".ssh/id_ed25519")),
 		wish.WithPublicKeyAuth(func(ctx ssh.Context, key ssh.PublicKey) bool {
-			return kc.IsAllowed(ctx, key)
+			return kc.IsAllowed(key)
 		}),
 		wish.WithMiddleware(
 			bm.Middleware(func(s ssh.Session) (tea.Model, []tea.ProgramOption) {
-				return tui.InitialModel(s.Context(), false, db, eng, version), []tea.ProgramOption{tea.WithAltScreen(), tea.WithMouseCellMotion()}
+				return tui.InitialModel(false, db, eng, version), []tea.ProgramOption{tea.WithAltScreen(), tea.WithMouseCellMotion()}
 			}),
 		),
 	)
@@ -509,7 +445,8 @@ func startSSHServer(port int, db store.Store, eng *monitor.Engine, kc *keyCache)
 	return s
 }
 
-func seedDemoData(ctx context.Context, s store.Store) {
+func seedDemoData(s store.Store) {
+	ctx := context.Background()
 	existing, _ := s.GetSites(ctx)
 	if len(existing) > 0 {
 		return
@@ -570,8 +507,8 @@ func newKeyCache(db store.Store) *keyCache {
 	return &keyCache{db: db, ttl: 30 * time.Second}
 }
 
-func (c *keyCache) refresh(ctx context.Context) {
-	users, err := c.db.GetAllUsers(ctx)
+func (c *keyCache) refresh() {
+	users, err := c.db.GetAllUsers(context.Background())
 	if err != nil {
 		// Keep the previous key set: a transient DB error must not lock every
 		// admin out. Revocation still fails closed because Invalidate clears
@@ -604,13 +541,13 @@ func (c *keyCache) Invalidate() {
 	c.mu.Unlock()
 }
 
-func (c *keyCache) IsAllowed(ctx context.Context, incomingKey ssh.PublicKey) bool {
+func (c *keyCache) IsAllowed(incomingKey ssh.PublicKey) bool {
 	c.mu.RLock()
 	stale := time.Since(c.updated) > c.ttl
 	c.mu.RUnlock()
 
 	if stale {
-		c.refresh(ctx)
+		c.refresh()
 	}
 
 	c.mu.RLock()
@@ -656,7 +593,8 @@ func (s *userInvalidatingStore) ImportData(ctx context.Context, data models.Back
 	return err
 }
 
-func seedKeysFromEnv(ctx context.Context, s store.Store) {
+func seedKeysFromEnv(s store.Store) {
+	ctx := context.Background()
 	var keys []string
 
 	if v := os.Getenv("UPTOP_ADMIN_KEY"); v != "" {

deploy/docker-compose.cluster.yml

@@ -3,10 +3,8 @@ services:
   # LEADER NODE
   # -------------------------
   leader:
-    image: lerkolabs/uptop:latest
+    build: .
     container_name: uptop-leader
-    sysctls:
-      - net.ipv4.ping_group_range=0 2147483647
     ports:
       - "23234:23234" # SSH
       - "8080:8080"   # HTTP
@@ -40,10 +38,8 @@ services:
   # FOLLOWER NODE
   # -------------------------
   follower:
-    image: lerkolabs/uptop:latest
+    build: .
     container_name: uptop-follower
-    sysctls:
-      - net.ipv4.ping_group_range=0 2147483647
     ports:
       - "23233:23234" # SSH (Mapped to different host port)
       - "8081:8080"   # HTTP (Mapped to different host port)

deploy/docker-compose.dev.yml

@@ -1,8 +1,8 @@
 services:
   # The Application
   app:
-    build:
-      context: ..
+    build: 
+      context: .
       dockerfile: Dockerfile
     container_name: uptop-dev
     ports:
@@ -13,7 +13,7 @@ services:
       - UPTOP_DB_TYPE=postgres
       - UPTOP_DB_DSN=postgres://devuser:devpass@postgres:5432/uptop_dev?sslmode=disable
       
-      # --- Web Server Configuration ---
+      # --- Web Server Configuration (Phase 4) ---
       - UPTOP_HTTP_PORT=8080
       - UPTOP_STATUS_ENABLED=true
       - UPTOP_STATUS_TITLE=Dev Infrastructure Status

deploy/docker-compose.probe.yml

@@ -1,8 +1,6 @@
 services:
   leader:
-    image: lerkolabs/uptop:latest
-    sysctls:
-      - net.ipv4.ping_group_range=0 2147483647
+    build: .
     environment:
       - UPTOP_CLUSTER_MODE=leader
       - UPTOP_CLUSTER_SECRET=changeme  # EXAMPLE ONLY — rotate before use
@@ -13,9 +11,7 @@ services:
       - "23234:23234"
 
   probe-us-east:
-    image: lerkolabs/uptop:latest
-    healthcheck:
-      disable: true
+    build: .
     environment:
       - UPTOP_CLUSTER_MODE=probe
       - UPTOP_NODE_ID=us-east-1
@@ -27,9 +23,7 @@ services:
       - leader
 
   probe-eu-west:
-    image: lerkolabs/uptop:latest
-    healthcheck:
-      disable: true
+    build: .
     environment:
       - UPTOP_CLUSTER_MODE=probe
       - UPTOP_NODE_ID=eu-west-1

deploy/docker-compose.yml

@@ -1,6 +1,8 @@
 services:
   app:
-    image: lerkolabs/uptop:latest
+    build:
+      context: .
+      dockerfile: Dockerfile
     container_name: uptop
     restart: unless-stopped
     read_only: true
@@ -8,8 +10,6 @@ services:
       - ALL
     security_opt:
       - no-new-privileges:true
-    sysctls:
-      - net.ipv4.ping_group_range=0 2147483647
     tmpfs:
       - /tmp
     ports:

docs/clustering.md

@@ -47,11 +47,13 @@ Probes are lightweight, stateless nodes that run checks from different locations
 | Node | Variable | Value |
 |------|----------|-------|
 | Both | `UPTOP_CLUSTER_SECRET` | Same shared secret |
+| Leader | `UPTOP_AGG_STRATEGY` | `any-down`, `majority-down`, or `all-down` |
 | Probe | `UPTOP_CLUSTER_MODE` | `probe` |
 | Probe | `UPTOP_PEER_URL` | Leader's HTTP URL |
 | Probe | `UPTOP_NODE_ID` | Unique identifier (e.g. `probe-us-east`) |
+| Probe | `UPTOP_NODE_REGION` | Region tag matching monitor assignments |
 
-Optional: `UPTOP_AGG_STRATEGY` (default `any-down`), `UPTOP_NODE_REGION` (omit to match all monitors), `UPTOP_NODE_NAME` (human-readable label in the TUI).
+Optional: `UPTOP_NODE_NAME` for a human-readable label in the TUI.
 
 See [`deploy/docker-compose.probe.yml`](../deploy/docker-compose.probe.yml) for a multi-region example.
 
@@ -78,6 +80,6 @@ Set via `UPTOP_AGG_STRATEGY` on the leader.
 
 ## Security
 
-- Set `UPTOP_CLUSTER_SECRET` on all nodes. Without it, cluster API endpoints reject all requests (fail closed); only `/api/health` stays open.
-- Secrets are sent in HTTP headers (`X-Uptop-Secret`). Use TLS or a reverse proxy for production.
+- Set `UPTOP_CLUSTER_SECRET` on all nodes. Without it, cluster API endpoints are unauthenticated.
+- Secrets are sent in HTTP headers (`X-Upkeep-Secret`). Use TLS or a reverse proxy for production.
 - uptop warns on startup if the cluster secret is missing or if cluster mode is active without TLS.

docs/config-as-code.md

@@ -122,7 +122,7 @@ Groups can't nest inside other groups. A group is healthy when all its children
 
 ## Alert types
 
-All 10 providers work in the YAML. The `settings` map is different per type.
+All 9 providers work in the YAML. The `settings` map is different per type.
 
 ```yaml
 # Discord / Slack / Generic Webhook — just a URL
@@ -149,9 +149,6 @@ All 10 providers work in the YAML. The `settings` map is different per type.
     url: https://ntfy.sh
     topic: my-alerts
     priority: "4"
-    # for protected topics:
-    # username: user
-    # password: pass
 
 # Telegram
 - name: Telegram Ops
@@ -181,14 +178,6 @@ All 10 providers work in the YAML. The `settings` map is different per type.
     url: https://gotify.example.com
     token: app-token
     priority: "8"
-
-# Opsgenie
-- name: Opsgenie
-  type: opsgenie
-  settings:
-    api_key: your-api-key
-    priority: P2        # P1–P5, default P3
-    # eu: "true"        # use the EU API endpoint
 ```
 
 ## Commands
@@ -235,25 +224,7 @@ Monitors and alerts are matched by **name**. Names must be unique across the ent
 
 Apply is idempotent. Run it twice with the same file, second run changes nothing.
 
-Apply is **not atomic** — items are written one at a time, so an error mid-apply (bad value, lost DB connection, ctrl-C) leaves the items already written in place. That's safe to recover from: apply diffs against the database by name, so fix the issue and run it again — it converges the rest. Just don't run two applies against the same database at once.
-
-## Backups and secrets
-
-`uptop export` writes alert credentials (SMTP passwords, API tokens, webhook URLs) into the YAML in clear text — that's what makes the file restorable. Treat it like a secrets file.
-
-The HTTP export endpoint redacts those same fields **by default**:
-
-```bash
-# secrets show as ***REDACTED*** — fine for sharing or review
-curl -H "X-Uptop-Secret: your-secret" \
-  "http://localhost:8080/api/backup/export"
-
-# full backup you can actually restore from
-curl -H "X-Uptop-Secret: your-secret" \
-  "http://localhost:8080/api/backup/export?redact_secrets=false"
-```
-
-Restoring a redacted export imports the literal string `***REDACTED***` as your credentials. For real backups, pass `redact_secrets=false` or run `uptop export` on the host.
+If something fails mid-apply, just fix the issue and run it again. It picks up where it left off.
 
 ## Typical workflow
 

go.mod

@@ -1,6 +1,6 @@
 module gitea.lerkolabs.com/lerkolabs/uptop
 
-go 1.26.5
+go 1.26.4
 
 require (
 	github.com/charmbracelet/bubbles v0.21.1-0.20250623103423-23b8fd6302d7

internal/alert/alert.go

@@ -18,9 +18,7 @@ import (
 	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
 )
 
-const alertHTTPTimeout = 10 * time.Second
-
-var alertClient = &http.Client{Timeout: alertHTTPTimeout}
+var alertClient = &http.Client{Timeout: 10 * time.Second}
 
 // sanitizeError strips the request URL from transport errors before they are
 // stored or displayed. *url.Error embeds the full URL, which for several

internal/cluster/cluster.go

@@ -38,34 +38,28 @@ func Start(ctx context.Context, cfg Config, eng *monitor.Engine) {
 	// "probe" mode is handled directly in main.go before cluster.Start is called
 }
 
-const (
-	followerTimeout        = 2 * time.Second
-	leaderFailureThreshold = 3
-	followerRetryInterval  = 5 * time.Second
-)
-
 func runFollowerLoop(ctx context.Context, cfg Config, eng *monitor.Engine) {
-	client := http.Client{Timeout: followerTimeout}
+	client := http.Client{Timeout: 2 * time.Second}
 	failures := 0
-	threshold := leaderFailureThreshold
+	threshold := 3
 
 	for {
 		select {
-		case <-time.After(followerRetryInterval):
+		case <-time.After(5 * time.Second):
 		case <-ctx.Done():
 			return
 		}
 
 		req, _ := http.NewRequest("GET", cfg.PeerURL+"/api/health", nil)
 		if cfg.SharedKey != "" {
-			req.Header.Set("X-Uptop-Secret", cfg.SharedKey)
+			req.Header.Set("X-Upkeep-Secret", cfg.SharedKey)
 		}
 
 		resp, err := client.Do(req)
 		isLeaderHealthy := false
 
 		if err == nil {
-			isLeaderHealthy = resp.StatusCode == http.StatusOK
+			isLeaderHealthy = resp.StatusCode == 200
 			_ = resp.Body.Close()
 		}
 

internal/cluster/cluster_test.go

@@ -113,7 +113,7 @@ func TestFollowerLoop_SendsSecret(t *testing.T) {
 	var receivedSecret string
 	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		mu.Lock()
-		receivedSecret = r.Header.Get("X-Uptop-Secret")
+		receivedSecret = r.Header.Get("X-Upkeep-Secret")
 		mu.Unlock()
 		w.WriteHeader(200)
 		w.Write([]byte("OK"))

internal/cluster/probe.go

@@ -26,18 +26,12 @@ type ProbeConfig struct {
 	AllowPrivateTargets bool
 }
 
-const (
-	probeMinInterval     = 10
-	probeDefaultInterval = 30
-	probeAPITimeout      = 10 * time.Second
-)
-
 func RunProbe(ctx context.Context, cfg ProbeConfig) error {
-	if cfg.Interval < probeMinInterval {
-		cfg.Interval = probeDefaultInterval
+	if cfg.Interval < 10 {
+		cfg.Interval = 30
 	}
 
-	apiClient := &http.Client{Timeout: probeAPITimeout}
+	apiClient := &http.Client{Timeout: 10 * time.Second}
 	dial := monitor.SafeDialContext(cfg.AllowPrivateTargets)
 	strictClient := &http.Client{
 		Transport: &http.Transport{
@@ -96,7 +90,7 @@ func probeRegister(ctx context.Context, client *http.Client, cfg ProbeConfig) er
 		return err
 	}
 	req.Header.Set("Content-Type", "application/json")
-	req.Header.Set("X-Uptop-Secret", cfg.SharedKey)
+	req.Header.Set("X-Upkeep-Secret", cfg.SharedKey)
 	resp, err := client.Do(req)
 	if err != nil {
 		return err
@@ -114,7 +108,7 @@ func probeFetchAssignments(ctx context.Context, client *http.Client, cfg ProbeCo
 	if err != nil {
 		return nil, err
 	}
-	req.Header.Set("X-Uptop-Secret", cfg.SharedKey)
+	req.Header.Set("X-Upkeep-Secret", cfg.SharedKey)
 	resp, err := client.Do(req)
 	if err != nil {
 		return nil, err
@@ -186,7 +180,7 @@ func probeReportResults(ctx context.Context, client *http.Client, cfg ProbeConfi
 		return err
 	}
 	req.Header.Set("Content-Type", "application/json")
-	req.Header.Set("X-Uptop-Secret", cfg.SharedKey)
+	req.Header.Set("X-Upkeep-Secret", cfg.SharedKey)
 	resp, err := client.Do(req)
 	if err != nil {
 		return err

internal/config/apply.go

@@ -54,7 +54,6 @@ func Apply(ctx context.Context, s store.Store, f *File, opts ApplyOpts) ([]Chang
 		alertMap[ea.Name] = ea.ID
 	}
 
-	nextPlaceholderID := -1
 	desiredAlertNames := make(map[string]bool, len(f.Alerts))
 	for _, a := range f.Alerts {
 		desiredAlertNames[a.Name] = true
@@ -67,9 +66,6 @@ func Apply(ctx context.Context, s store.Store, f *File, opts ApplyOpts) ([]Chang
 					return changes, fmt.Errorf("create alert %q: %w", a.Name, err)
 				}
 				alertMap[a.Name] = id
-			} else {
-				alertMap[a.Name] = nextPlaceholderID
-				nextPlaceholderID--
 			}
 		} else {
 			alertMap[a.Name] = existing.ID
@@ -113,9 +109,6 @@ func Apply(ctx context.Context, s store.Store, f *File, opts ApplyOpts) ([]Chang
 					return changes, fmt.Errorf("create group %q: %w", g.Name, err)
 				}
 				groupMap[g.Name] = id
-			} else {
-				groupMap[g.Name] = nextPlaceholderID
-				nextPlaceholderID--
 			}
 		} else {
 			groupMap[g.Name] = existing.ID

internal/config/apply_test.go

@@ -266,74 +266,6 @@ func TestApplyDuplicateNames(t *testing.T) {
 	}
 }
 
-func TestApplyDryRunNewAlertAndMonitor(t *testing.T) {
-	s := newTestStore(t)
-	f := &File{
-		Alerts: []Alert{
-			{Name: "Discord", Type: "discord", Settings: map[string]string{"url": "https://example.com"}},
-		},
-		Monitors: []Monitor{
-			{Name: "Web", Type: "http", URL: "https://example.com", Interval: 30, Alert: "Discord"},
-		},
-	}
-
-	changes, err := Apply(context.Background(), s, f, ApplyOpts{DryRun: true})
-	if err != nil {
-		t.Fatalf("dry-run with new alert+monitor should not error: %v", err)
-	}
-
-	creates := 0
-	for _, c := range changes {
-		if c.Action == "create" {
-			creates++
-		}
-	}
-	if creates != 2 {
-		t.Fatalf("expected 2 creates (alert+monitor), got %d: %+v", creates, changes)
-	}
-
-	sites, _ := s.GetSites(context.Background())
-	alerts, _ := s.GetAllAlerts(context.Background())
-	if len(sites) != 0 {
-		t.Fatalf("dry-run should not persist sites, got %d", len(sites))
-	}
-	if len(alerts) != 0 {
-		t.Fatalf("dry-run should not persist alerts, got %d", len(alerts))
-	}
-}
-
-func TestApplyDryRunNewGroupWithChildren(t *testing.T) {
-	s := newTestStore(t)
-	f := &File{
-		Alerts: []Alert{
-			{Name: "Slack", Type: "slack", Settings: map[string]string{"url": "https://hooks.example.com"}},
-		},
-		Monitors: []Monitor{
-			{
-				Name: "Prod", Type: "group", Alert: "Slack",
-				Monitors: []Monitor{
-					{Name: "API", Type: "http", URL: "https://api.example.com", Interval: 15, Alert: "Slack"},
-				},
-			},
-		},
-	}
-
-	changes, err := Apply(context.Background(), s, f, ApplyOpts{DryRun: true})
-	if err != nil {
-		t.Fatalf("dry-run with new group+alert should not error: %v", err)
-	}
-
-	creates := 0
-	for _, c := range changes {
-		if c.Action == "create" {
-			creates++
-		}
-	}
-	if creates != 3 {
-		t.Fatalf("expected 3 creates (alert+group+child), got %d: %+v", creates, changes)
-	}
-}
-
 func TestApplyExistingAlertReference(t *testing.T) {
 	s := newTestStore(t)
 	s.AddAlert(context.Background(), "Existing", "webhook", map[string]string{"url": "https://example.com"})

internal/models/models.go

@@ -86,11 +86,6 @@ type ProbeNode struct {
 	Version  string
 }
 
-type LogEntry struct {
-	Message   string
-	CreatedAt time.Time
-}
-
 // AlertHealthRecord is the persisted send health of an alert channel. It lets the
 // "last sent" / health indicators survive restarts instead of resetting to "never".
 type AlertHealthRecord struct {

internal/monitor/alerts.go

@@ -1,272 +0,0 @@
-package monitor
-
-import (
-	"context"
-	"fmt"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/alert"
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-type AlertHealth struct {
-	LastSendAt time.Time
-	LastSendOK bool
-	LastError  string
-	SendCount  int
-	FailCount  int
-}
-
-// InitAlertHealth restores persisted alert send health so the dashboard shows real
-// "last sent" / health state on startup instead of resetting every channel to "never".
-func (e *Engine) InitAlertHealth(ctx context.Context) {
-	records, err := e.db.LoadAlertHealth(ctx)
-	if err != nil {
-		return
-	}
-	e.alertHealthMu.Lock()
-	defer e.alertHealthMu.Unlock()
-	for id, r := range records {
-		e.alertHealth[id] = AlertHealth{
-			LastSendAt: r.LastSendAt,
-			LastSendOK: r.LastSendOK,
-			LastError:  r.LastError,
-			SendCount:  r.SendCount,
-			FailCount:  r.FailCount,
-		}
-	}
-}
-
-// handleStatusChange folds a check result into the live state. snap is the
-// stale snapshot the check ran against; the actual mutation is applied onto the
-// CURRENT live entry via applyState, so a concurrent pause / config edit /
-// heartbeat is never reverted by this write. Logs and alerts are emitted after
-// the lock is released, off the critical section.
-func (e *Engine) handleStatusChange(snap models.Site, rawStatus string, code int, latency time.Duration, errorReason string) {
-	if !e.IsActive() {
-		return
-	}
-
-	inMaint := e.isInMaintenance(snap.ID)
-	status := models.Status(rawStatus)
-
-	var (
-		prev, next            models.Status
-		name, typ             string
-		alertID               int
-		failCount, maxRetries int
-		confirmedDown         bool
-		failedCheck           bool
-		downSince             time.Time
-		sslWarnFire           bool
-		sslDays               int
-		skipped               bool
-		changed               bool
-	)
-
-	_, exists := e.applyState(snap.ID, func(s *models.Site) {
-		// A non-UP result computed from a stale snapshot must not override a
-		// heartbeat (or newer check) that landed while we were evaluating.
-		if status != models.StatusUp && s.LastCheck.After(snap.LastCheck) {
-			skipped = true
-			return
-		}
-
-		prev = s.Status
-		name = s.Name
-		typ = s.Type
-		alertID = s.AlertID
-		maxRetries = s.MaxRetries
-		downSince = s.StatusChangedAt
-
-		// Fresh check results (measured by the run against snap).
-		s.StatusCode = code
-		s.Latency = snap.Latency
-		s.LastCheck = snap.LastCheck
-		s.HasSSL = snap.HasSSL
-		s.CertExpiry = snap.CertExpiry
-		s.LastError = errorReason
-		if status == models.StatusUp {
-			s.LastSuccessAt = time.Now()
-			s.LastError = ""
-		}
-
-		// Status + failure-count transition, based on the CURRENT live status.
-		if status == models.StatusUp {
-			s.FailureCount = 0
-			s.Status = models.StatusUp
-		} else {
-			if s.FailureCount <= s.MaxRetries {
-				s.FailureCount++
-			}
-			if s.FailureCount > s.MaxRetries {
-				if s.Status != status {
-					confirmedDown = true
-				}
-				s.Status = status
-				s.FailureCount = s.MaxRetries + 1
-			} else {
-				failedCheck = true
-			}
-		}
-		failCount = s.FailureCount
-
-		if s.Status != prev && prev != models.StatusPending {
-			s.StatusChangedAt = time.Now()
-		} else if s.StatusChangedAt.IsZero() && s.Status != models.StatusPending {
-			s.StatusChangedAt = time.Now()
-		}
-
-		// SSL expiry warning (fresh HasSSL/CertExpiry + config threshold).
-		if typ == "http" && s.CheckSSL && s.HasSSL {
-			days := int(time.Until(s.CertExpiry).Hours() / 24)
-			if days <= s.ExpiryThreshold && !s.SentSSLWarning && status != models.StatusSSLExp {
-				sslWarnFire = true
-				sslDays = days
-				s.SentSSLWarning = true
-			} else if days > s.ExpiryThreshold {
-				s.SentSSLWarning = false
-			}
-		}
-
-		next = s.Status
-		changed = next != prev
-	})
-
-	if !exists || skipped {
-		return
-	}
-
-	e.recordCheck(snap.ID, latency, status == models.StatusUp)
-
-	if confirmedDown {
-		if errorReason != "" {
-			e.AddLog(fmt.Sprintf("Monitor '%s' confirmed DOWN: %s", name, errorReason))
-		} else {
-			e.AddLog(fmt.Sprintf("Monitor '%s' confirmed DOWN", name))
-		}
-	} else if failedCheck {
-		e.AddLog(fmt.Sprintf("Monitor '%s' failed check %d/%d", name, failCount, maxRetries))
-	}
-
-	if changed && prev != models.StatusPending {
-		e.enqueueWrite(writeStateChange{siteID: snap.ID, fromStatus: string(prev), toStatus: string(next), reason: errorReason})
-	}
-
-	if sslWarnFire {
-		if !inMaint {
-			e.triggerAlert(alertID, "SSL WARNING", fmt.Sprintf("SSL for '%s' expires in %d days", name, sslDays))
-		} else {
-			e.AddLog(fmt.Sprintf("SSL warning for '%s' suppressed (maintenance)", name))
-		}
-	}
-
-	if prev == models.StatusUp && next == models.StatusLate {
-		e.AddLog(fmt.Sprintf("Monitor '%s' heartbeat overdue", name))
-	}
-
-	if !prev.IsBroken() && next.IsBroken() && next != models.StatusPending {
-		if inMaint {
-			e.AddLog(fmt.Sprintf("Monitor '%s' is DOWN (alerts suppressed — maintenance)", name))
-		} else {
-			msg := fmt.Sprintf("Monitor '%s' is DOWN (%s)", name, rawStatus)
-			if errorReason != "" {
-				msg = fmt.Sprintf("Monitor '%s' is DOWN: %s", name, errorReason)
-			}
-			if typ == "push" {
-				msg = fmt.Sprintf("Push Monitor '%s' missed heartbeat.", name)
-			}
-			e.triggerAlert(alertID, "🚨 ALERT", msg)
-		}
-	}
-	if prev.IsBroken() && next == models.StatusUp {
-		downDur := ""
-		if !downSince.IsZero() {
-			downDur = fmt.Sprintf(" (was down %s)", fmtDurationShort(time.Since(downSince)))
-		}
-		e.AddLog(fmt.Sprintf("Monitor '%s' recovered%s", name, downDur))
-		if !inMaint {
-			e.triggerAlert(alertID, "✅ RECOVERY", fmt.Sprintf("Monitor '%s' is UP%s", name, downDur))
-		}
-	}
-	if prev == models.StatusLate && next == models.StatusUp && !prev.IsBroken() {
-		e.AddLog(fmt.Sprintf("Monitor '%s' heartbeat arrived (was late)", name))
-	}
-}
-
-func (e *Engine) triggerAlert(alertID int, title, message string) {
-	if alertID <= 0 {
-		return
-	}
-	cfg, err := e.db.GetAlert(e.ctx, alertID)
-	if err != nil {
-		e.AddLog(fmt.Sprintf("Failed to load alert config %d: %v", alertID, err))
-		return
-	}
-	provider := alert.GetProvider(cfg)
-	if provider != nil {
-		go func() {
-			ctx, cancel := context.WithTimeout(e.ctx, alertSendTimeout)
-			defer cancel()
-			if err := provider.Send(ctx, title, message); err != nil {
-				e.AddLog(fmt.Sprintf("Alert send failed (%s): %v", cfg.Name, err))
-				e.recordAlertResult(alertID, false, err.Error())
-			} else {
-				e.recordAlertResult(alertID, true, "")
-			}
-		}()
-	}
-}
-
-func (e *Engine) recordAlertResult(alertID int, ok bool, errMsg string) {
-	e.alertHealthMu.Lock()
-	defer e.alertHealthMu.Unlock()
-	h := e.alertHealth[alertID]
-	h.LastSendAt = time.Now()
-	h.LastSendOK = ok
-	h.SendCount++
-	if ok {
-		h.LastError = ""
-	} else {
-		h.LastError = errMsg
-		h.FailCount++
-	}
-	e.alertHealth[alertID] = h
-
-	// Persist so health survives restarts; DB IO off the alert path.
-	e.enqueueWrite(writeAlertHealth{rec: models.AlertHealthRecord{
-		AlertID:    alertID,
-		LastSendAt: h.LastSendAt,
-		LastSendOK: h.LastSendOK,
-		LastError:  h.LastError,
-		SendCount:  h.SendCount,
-		FailCount:  h.FailCount,
-	}})
-}
-
-func (e *Engine) GetAlertHealth(alertID int) AlertHealth {
-	e.alertHealthMu.RLock()
-	defer e.alertHealthMu.RUnlock()
-	return e.alertHealth[alertID]
-}
-
-func (e *Engine) TestAlert(ctx context.Context, alertID int) error {
-	cfg, err := e.db.GetAlert(ctx, alertID)
-	if err != nil {
-		return fmt.Errorf("failed to load alert: %w", err)
-	}
-	provider := alert.GetProvider(cfg)
-	if provider == nil {
-		return fmt.Errorf("no provider for type %q", cfg.Type)
-	}
-	ctx, cancel := context.WithTimeout(ctx, alertSendTimeout)
-	defer cancel()
-	err = provider.Send(ctx, "🧪 Test Alert", fmt.Sprintf("Test notification from uptop for channel '%s'.", cfg.Name))
-	if err != nil {
-		e.recordAlertResult(alertID, false, err.Error())
-		return err
-	}
-	e.recordAlertResult(alertID, true, "")
-	e.AddLog(fmt.Sprintf("Test alert sent to '%s'", cfg.Name))
-	return nil
-}

internal/monitor/alerts_test.go

@@ -1,487 +0,0 @@
-package monitor
-
-import (
-	"context"
-	"testing"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-// --- Group 1: State Machine ---
-
-func TestHandleStatusChange_PendingToUp(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 3, AlertID: 1},
-		SiteState:  models.SiteState{Status: "PENDING"},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "UP", 200, 10*time.Millisecond, "")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected UP, got %s", s.Status)
-	}
-	if s.FailureCount != 0 {
-		t.Errorf("expected FailureCount 0, got %d", s.FailureCount)
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) != 0 {
-		t.Error("expected no alert for PENDING→UP")
-	}
-}
-
-func TestHandleStatusChange_UpIncrementFailure(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 3},
-		SiteState:  models.SiteState{Status: "UP", FailureCount: 0},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "DOWN", 500, 0, "test error")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected UP (under retry threshold), got %s", s.Status)
-	}
-	if s.FailureCount != 1 {
-		t.Errorf("expected FailureCount 1, got %d", s.FailureCount)
-	}
-}
-
-func TestHandleStatusChange_UpToDown_ExceedsRetries(t *testing.T) {
-	ms := newMockStore()
-	ms.alerts[1] = models.AlertConfig{ID: 1, Name: "discord", Type: "webhook", Settings: map[string]string{"url": "http://example.com"}}
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 2, AlertID: 1},
-		SiteState:  models.SiteState{Status: "UP", FailureCount: 2},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "DOWN", 500, 0, "test error")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "DOWN" {
-		t.Errorf("expected DOWN, got %s", s.Status)
-	}
-	if s.FailureCount != 3 {
-		t.Errorf("expected FailureCount 3, got %d", s.FailureCount)
-	}
-	waitAsync()
-	calls := ms.getAlertCallsSnapshot()
-	if len(calls) == 0 || calls[0] != 1 {
-		t.Errorf("expected alert call for alertID 1, got %v", calls)
-	}
-}
-
-func TestHandleStatusChange_UpToDown_ZeroRetries(t *testing.T) {
-	ms := newMockStore()
-	ms.alerts[1] = models.AlertConfig{ID: 1, Name: "test", Type: "webhook", Settings: map[string]string{"url": "http://example.com"}}
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 0, AlertID: 1},
-		SiteState:  models.SiteState{Status: "UP", FailureCount: 0},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "DOWN", 0, 0, "test error")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "DOWN" {
-		t.Errorf("expected DOWN, got %s", s.Status)
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) == 0 {
-		t.Error("expected alert on immediate DOWN")
-	}
-}
-
-func TestHandleStatusChange_DownToUp_Recovery(t *testing.T) {
-	ms := newMockStore()
-	ms.alerts[1] = models.AlertConfig{ID: 1, Name: "test", Type: "webhook", Settings: map[string]string{"url": "http://example.com"}}
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", AlertID: 1},
-		SiteState:  models.SiteState{Status: "DOWN", FailureCount: 4},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "UP", 200, 5*time.Millisecond, "")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected UP, got %s", s.Status)
-	}
-	if s.FailureCount != 0 {
-		t.Errorf("expected FailureCount 0, got %d", s.FailureCount)
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) == 0 {
-		t.Error("expected recovery alert")
-	}
-}
-
-func TestHandleStatusChange_DownStaysDown(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 2},
-		SiteState:  models.SiteState{Status: "DOWN", FailureCount: 3},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "DOWN", 0, 0, "test error")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "DOWN" {
-		t.Errorf("expected DOWN, got %s", s.Status)
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) != 0 {
-		t.Error("expected no re-alert for already DOWN")
-	}
-}
-
-func TestHandleStatusChange_SSLExpired(t *testing.T) {
-	ms := newMockStore()
-	ms.alerts[1] = models.AlertConfig{ID: 1, Name: "test", Type: "webhook", Settings: map[string]string{"url": "http://example.com"}}
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 0, AlertID: 1},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "SSL EXP", 0, 0, "SSL certificate expired")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "SSL EXP" {
-		t.Errorf("expected SSL EXP, got %s", s.Status)
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) == 0 {
-		t.Error("expected alert on SSL EXP")
-	}
-}
-
-func TestHandleStatusChange_AlertSuppressedMaintenance(t *testing.T) {
-	ms := newMockStore()
-	ms.maintenance[1] = true
-	ms.alerts[1] = models.AlertConfig{ID: 1, Name: "test", Type: "webhook", Settings: map[string]string{"url": "http://example.com"}}
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 0, AlertID: 1},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-	e.refreshMaintenanceCache(context.Background())
-
-	e.handleStatusChange(site, "DOWN", 0, 0, "test error")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "DOWN" {
-		t.Errorf("expected DOWN, got %s", s.Status)
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) != 0 {
-		t.Error("expected no alert during maintenance")
-	}
-	logs := e.GetLogs()
-	found := false
-	for _, l := range logs {
-		if containsStr(l.Message, "suppressed") {
-			found = true
-			break
-		}
-	}
-	if !found {
-		t.Error("expected log mentioning suppressed")
-	}
-}
-
-func TestHandleStatusChange_RecoverySuppressedMaintenance(t *testing.T) {
-	ms := newMockStore()
-	ms.maintenance[1] = true
-	ms.alerts[1] = models.AlertConfig{ID: 1, Name: "test", Type: "webhook", Settings: map[string]string{"url": "http://example.com"}}
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", AlertID: 1},
-		SiteState:  models.SiteState{Status: "DOWN"},
-	}
-	injectSite(e, site)
-	e.refreshMaintenanceCache(context.Background())
-
-	e.handleStatusChange(site, "UP", 200, 0, "")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected UP, got %s", s.Status)
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) != 0 {
-		t.Error("expected no alert during maintenance recovery")
-	}
-}
-
-func TestHandleStatusChange_SSLWarning(t *testing.T) {
-	ms := newMockStore()
-	ms.alerts[1] = models.AlertConfig{ID: 1, Name: "test", Type: "webhook", Settings: map[string]string{"url": "http://example.com"}}
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", Type: "http", CheckSSL: true, ExpiryThreshold: 30, AlertID: 1},
-		SiteState:  models.SiteState{Status: "UP", HasSSL: true, SentSSLWarning: false, CertExpiry: time.Now().Add(15 * 24 * time.Hour)},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "UP", 200, 0, "")
-
-	s, _ := getSite(e, 1)
-	if !s.SentSSLWarning {
-		t.Error("expected SentSSLWarning=true")
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) == 0 {
-		t.Error("expected SSL warning alert")
-	}
-}
-
-func TestHandleStatusChange_SSLWarningNotRepeated(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", Type: "http", CheckSSL: true, ExpiryThreshold: 30, AlertID: 1},
-		SiteState:  models.SiteState{Status: "UP", HasSSL: true, SentSSLWarning: true, CertExpiry: time.Now().Add(15 * 24 * time.Hour)},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "UP", 200, 0, "")
-
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) != 0 {
-		t.Error("expected no repeat SSL warning")
-	}
-}
-
-func TestHandleStatusChange_SSLWarningReset(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", Type: "http", CheckSSL: true, ExpiryThreshold: 30},
-		SiteState:  models.SiteState{Status: "UP", HasSSL: true, SentSSLWarning: true, CertExpiry: time.Now().Add(60 * 24 * time.Hour)},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "UP", 200, 0, "")
-
-	s, _ := getSite(e, 1)
-	if s.SentSSLWarning {
-		t.Error("expected SentSSLWarning reset to false")
-	}
-}
-
-func TestHandleStatusChange_SSLWarningSuppressedMaint(t *testing.T) {
-	ms := newMockStore()
-	ms.maintenance[1] = true
-	ms.alerts[1] = models.AlertConfig{ID: 1, Name: "test", Type: "webhook", Settings: map[string]string{"url": "http://example.com"}}
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", Type: "http", CheckSSL: true, ExpiryThreshold: 30, AlertID: 1},
-		SiteState:  models.SiteState{Status: "UP", HasSSL: true, SentSSLWarning: false, CertExpiry: time.Now().Add(15 * 24 * time.Hour)},
-	}
-	injectSite(e, site)
-	e.refreshMaintenanceCache(context.Background())
-
-	e.handleStatusChange(site, "UP", 200, 0, "")
-
-	s, _ := getSite(e, 1)
-	if !s.SentSSLWarning {
-		t.Error("expected SentSSLWarning=true even in maintenance")
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) != 0 {
-		t.Error("expected no alert during maintenance")
-	}
-}
-
-func TestHandleStatusChange_InactiveEngine(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 0},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-	e.SetActive(false)
-
-	e.handleStatusChange(site, "DOWN", 0, 0, "test error")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Error("expected no state change when inactive")
-	}
-}
-
-// --- Group 10: liveState merge (lost-update race) ---
-
-// A pause that lands while a check is in flight must survive the check's
-// write-back. The old code snapshotted the site, ran the check, then wrote the
-// whole stale struct back — reverting the pause.
-func TestHandleStatusChange_PauseDuringCheckSurvives(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 0},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-
-	// `site` is the stale snapshot the check ran against (Paused=false).
-	// Meanwhile the user pauses the monitor.
-	e.ToggleSitePause(1)
-
-	// Check completes and folds its result in using the stale snapshot.
-	e.handleStatusChange(site, "DOWN", 500, 0, "boom")
-
-	s, _ := getSite(e, 1)
-	if !s.Paused {
-		t.Error("pause was reverted by a stale check write-back")
-	}
-	if s.Status != "DOWN" {
-		t.Errorf("expected check result still applied (DOWN), got %s", s.Status)
-	}
-}
-
-// A config edit that lands while a check is in flight must survive; the check
-// must not resurrect the old config from its snapshot.
-func TestHandleStatusChange_ConfigEditDuringCheckSurvives(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", URL: "http://old.com", Type: "http", MaxRetries: 0, Interval: 30},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-
-	// Config changes mid-check.
-	e.UpdateSiteConfig(models.SiteConfig{ID: 1, Name: "test", URL: "http://new.com", Type: "http", Interval: 60})
-
-	// Stale check (ran against http://old.com) folds its result in.
-	e.handleStatusChange(site, "UP", 200, 5*time.Millisecond, "")
-
-	s, _ := getSite(e, 1)
-	if s.URL != "http://new.com" {
-		t.Errorf("config edit reverted: URL=%s", s.URL)
-	}
-	if s.Interval != 60 {
-		t.Errorf("config edit reverted: Interval=%d", s.Interval)
-	}
-}
-
-// The classic push false-DOWN: a heartbeat marks the monitor UP while a
-// staleness evaluation (computed from the older LastCheck) is mid-flight.
-// The stale DOWN must not overwrite the fresh heartbeat.
-func TestHandleStatusChange_HeartbeatNotOverwrittenByStaleDown(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	// Snapshot the engine would have taken before evaluating staleness:
-	// LastCheck is old, so checkPush decided "DOWN".
-	snap := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "push", Type: "push", Token: "tok", Interval: 10},
-		SiteState:  models.SiteState{Status: "UP", LastCheck: time.Now().Add(-120 * time.Second)},
-	}
-	injectSite(e, snap)
-
-	// A heartbeat lands first, advancing LastCheck and confirming UP.
-	if !e.RecordHeartbeat("tok") {
-		t.Fatal("heartbeat rejected")
-	}
-
-	// Now the in-flight stale evaluation tries to write DOWN.
-	e.handleStatusChange(snap, "DOWN", 0, 0, "heartbeat missed")
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("stale DOWN overwrote a fresh heartbeat: status=%s", s.Status)
-	}
-}
-
-// A check result for a site removed mid-check must be dropped, not recreate it.
-func TestHandleStatusChange_RemovedSiteDropped(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", MaxRetries: 0},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-
-	e.RemoveSite(1)
-	e.handleStatusChange(site, "DOWN", 500, 0, "boom")
-
-	if _, ok := getSite(e, 1); ok {
-		t.Error("removed site was recreated by a late check write-back")
-	}
-}
-
-// --- Group 12: Phase 3 engine correctness ---
-
-// PENDING→DOWN must honor MaxRetries instead of alerting on first failure.
-func TestHandleStatusChange_PendingRetriesBeforeDown(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "new-monitor", MaxRetries: 2},
-		SiteState:  models.SiteState{Status: "PENDING"},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "DOWN", 0, 0, "timeout")
-	s, _ := getSite(e, 1)
-	if s.Status != "PENDING" {
-		t.Errorf("expected PENDING during retry, got %s", s.Status)
-	}
-	if s.FailureCount != 1 {
-		t.Errorf("expected FailureCount 1, got %d", s.FailureCount)
-	}
-
-	e.handleStatusChange(s, "DOWN", 0, 0, "timeout")
-	s, _ = getSite(e, 1)
-	if s.Status != "PENDING" {
-		t.Errorf("expected PENDING during retry 2, got %s", s.Status)
-	}
-
-	e.handleStatusChange(s, "DOWN", 0, 0, "timeout")
-	s, _ = getSite(e, 1)
-	if s.Status != "DOWN" {
-		t.Errorf("expected DOWN after retries exhausted, got %s", s.Status)
-	}
-}
-
-// LATE→DOWN must also honor MaxRetries.
-func TestHandleStatusChange_LateRetriesBeforeDown(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "push-mon", MaxRetries: 1},
-		SiteState:  models.SiteState{Status: "LATE"},
-	}
-	injectSite(e, site)
-
-	e.handleStatusChange(site, "DOWN", 0, 0, "missed heartbeat")
-	s, _ := getSite(e, 1)
-	if s.Status != "LATE" {
-		t.Errorf("expected LATE during retry, got %s", s.Status)
-	}
-
-	e.handleStatusChange(s, "DOWN", 0, 0, "missed heartbeat")
-	s, _ = getSite(e, 1)
-	if s.Status != "DOWN" {
-		t.Errorf("expected DOWN after retries exhausted, got %s", s.Status)
-	}
-}

internal/monitor/checker_test.go

@@ -130,14 +130,8 @@ func TestRunCheck_Port_Open(t *testing.T) {
 	}
 	defer ln.Close()
 
-	_, portStr, err := net.SplitHostPort(ln.Addr().String())
-	if err != nil {
-		t.Fatalf("SplitHostPort: %v", err)
-	}
-	port, err := strconv.Atoi(portStr)
-	if err != nil {
-		t.Fatalf("Atoi: %v", err)
-	}
+	_, portStr, _ := net.SplitHostPort(ln.Addr().String())
+	port, _ := strconv.Atoi(portStr)
 
 	site := models.SiteConfig{ID: 1, Type: "port", Hostname: "127.0.0.1", Port: port, Timeout: 2}
 	result := RunCheck(context.Background(), site, nil, nil, false, true)
@@ -155,14 +149,8 @@ func TestRunCheck_Port_Closed(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	_, portStr, err := net.SplitHostPort(ln.Addr().String())
-	if err != nil {
-		t.Fatalf("SplitHostPort: %v", err)
-	}
-	port, err := strconv.Atoi(portStr)
-	if err != nil {
-		t.Fatalf("Atoi: %v", err)
-	}
+	_, portStr, _ := net.SplitHostPort(ln.Addr().String())
+	port, _ := strconv.Atoi(portStr)
 	ln.Close()
 
 	site := models.SiteConfig{ID: 1, Type: "port", Hostname: "127.0.0.1", Port: port, Timeout: 1}
@@ -180,14 +168,8 @@ func TestRunPortCheck_UsesPinnedIP(t *testing.T) {
 	}
 	defer ln.Close()
 
-	_, portStr, err := net.SplitHostPort(ln.Addr().String())
-	if err != nil {
-		t.Fatalf("SplitHostPort: %v", err)
-	}
-	port, err := strconv.Atoi(portStr)
-	if err != nil {
-		t.Fatalf("Atoi: %v", err)
-	}
+	_, portStr, _ := net.SplitHostPort(ln.Addr().String())
+	port, _ := strconv.Atoi(portStr)
 
 	// Pass a pinned IP — runPortCheck should dial it instead of resolving Hostname.
 	site := models.SiteConfig{ID: 1, Type: "port", Hostname: "will-not-resolve.invalid", Port: port, Timeout: 2}
@@ -205,14 +187,8 @@ func TestRunPortCheck_NilPinnedIP_UsesHostname(t *testing.T) {
 	}
 	defer ln.Close()
 
-	_, portStr, err := net.SplitHostPort(ln.Addr().String())
-	if err != nil {
-		t.Fatalf("SplitHostPort: %v", err)
-	}
-	port, err := strconv.Atoi(portStr)
-	if err != nil {
-		t.Fatalf("Atoi: %v", err)
-	}
+	_, portStr, _ := net.SplitHostPort(ln.Addr().String())
+	port, _ := strconv.Atoi(portStr)
 
 	site := models.SiteConfig{ID: 1, Type: "port", Hostname: "127.0.0.1", Port: port, Timeout: 2}
 	result := runPortCheck(context.Background(), site, nil)
@@ -229,14 +205,8 @@ func TestRunCheck_Port_BlocksPrivateByDefault(t *testing.T) {
 	}
 	defer ln.Close()
 
-	_, portStr, err := net.SplitHostPort(ln.Addr().String())
-	if err != nil {
-		t.Fatalf("SplitHostPort: %v", err)
-	}
-	port, err := strconv.Atoi(portStr)
-	if err != nil {
-		t.Fatalf("Atoi: %v", err)
-	}
+	_, portStr, _ := net.SplitHostPort(ln.Addr().String())
+	port, _ := strconv.Atoi(portStr)
 
 	site := models.SiteConfig{ID: 1, Type: "port", Hostname: "127.0.0.1", Port: port, Timeout: 2}
 	result := RunCheck(context.Background(), site, nil, nil, false, false)

internal/monitor/checks.go

@@ -1,329 +0,0 @@
-package monitor
-
-import (
-	"context"
-	"fmt"
-	"math/rand/v2"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-func (e *Engine) RecordHeartbeat(token string) bool {
-	if !e.IsActive() {
-		return false
-	}
-
-	e.mu.RLock()
-	targetID, ok := e.tokenIndex[token]
-	e.mu.RUnlock()
-	if !ok {
-		return false
-	}
-
-	var (
-		prevStatus models.Status
-		name       string
-		alertID    int
-		downSince  time.Time
-	)
-	_, exists := e.applyState(targetID, func(s *models.Site) {
-		prevStatus = s.Status
-		name = s.Name
-		alertID = s.AlertID
-		downSince = s.StatusChangedAt // captured before mutation = when it went down
-
-		s.LastCheck = time.Now()
-		s.Status = models.StatusUp
-		s.FailureCount = 0
-		s.Latency = 0
-		s.LastError = ""
-		s.LastSuccessAt = time.Now()
-		if prevStatus != models.StatusUp {
-			s.StatusChangedAt = time.Now()
-		}
-	})
-	if !exists {
-		return false
-	}
-
-	switch prevStatus {
-	case models.StatusPending:
-		e.AddLog(fmt.Sprintf("Push Monitor '%s' received first heartbeat", name))
-	case models.StatusLate:
-		e.AddLog(fmt.Sprintf("Push Monitor '%s' heartbeat arrived (was late)", name))
-	case models.StatusStale:
-		e.AddLog(fmt.Sprintf("Push Monitor '%s' heartbeat arrived (was stale)", name))
-	case models.StatusDown:
-		downDur := ""
-		if !downSince.IsZero() {
-			downDur = fmt.Sprintf(" (was down %s)", fmtDurationShort(time.Since(downSince)))
-		}
-		e.AddLog(fmt.Sprintf("Push Monitor '%s' recovered%s", name, downDur))
-		go e.triggerAlert(alertID, "✅ RECOVERY", fmt.Sprintf("Push Monitor '%s' is receiving heartbeats.%s", name, downDur))
-	}
-
-	e.recordCheck(targetID, 0, true)
-
-	if prevStatus != models.StatusUp && prevStatus != models.StatusPending {
-		e.enqueueWrite(writeStateChange{siteID: targetID, fromStatus: string(prevStatus), toStatus: string(models.StatusUp)})
-	}
-
-	return true
-}
-
-func (e *Engine) getRecheckChan(id int) chan struct{} {
-	e.recheckMu.Lock()
-	defer e.recheckMu.Unlock()
-	ch, ok := e.recheck[id]
-	if !ok {
-		ch = make(chan struct{}, 1)
-		e.recheck[id] = ch
-	}
-	return ch
-}
-
-func (e *Engine) signalRecheck(id int) {
-	ch := e.getRecheckChan(id)
-	select {
-	case ch <- struct{}{}:
-	default:
-	}
-}
-
-func (e *Engine) monitorRoutine(ctx context.Context, id int) {
-	recheckCh := e.getRecheckChan(id)
-
-	// Stagger initial check to avoid thundering herd on startup
-	stagger := time.Duration(rand.IntN(3000)) * time.Millisecond //nolint:gosec // non-security jitter
-	select {
-	case <-time.After(stagger):
-	case <-ctx.Done():
-		return
-	}
-
-	e.checkByID(ctx, id)
-	for {
-		select {
-		case <-ctx.Done():
-			return
-		default:
-		}
-
-		if !e.IsActive() {
-			select {
-			case <-time.After(pollInterval):
-			case <-ctx.Done():
-				return
-			case <-recheckCh:
-			}
-			continue
-		}
-
-		e.mu.RLock()
-		site, exists := e.liveState[id]
-		e.mu.RUnlock()
-		if !exists {
-			return
-		}
-
-		if site.Paused {
-			select {
-			case <-time.After(pollInterval):
-			case <-ctx.Done():
-				return
-			case <-recheckCh:
-			}
-			continue
-		}
-
-		interval := site.Interval
-		if interval < minCheckInterval {
-			interval = minCheckInterval
-		}
-		jitter := time.Duration(rand.IntN(interval*100)) * time.Millisecond //nolint:gosec // non-security jitter
-		select {
-		case <-time.After(time.Duration(interval)*time.Second + jitter):
-		case <-ctx.Done():
-			return
-		case <-recheckCh:
-		}
-		e.checkByID(ctx, id)
-	}
-}
-
-func (e *Engine) checkByID(ctx context.Context, id int) {
-	if !e.IsActive() {
-		return
-	}
-
-	e.mu.RLock()
-	site, exists := e.liveState[id]
-	e.mu.RUnlock()
-	if !exists || site.Paused {
-		return
-	}
-
-	switch site.Type {
-	case "push":
-		e.checkPush(ctx, site)
-	case "group":
-		e.checkGroup(ctx, site)
-	default:
-		result := RunCheck(ctx, site.SiteConfig, e.strictClient, e.insecureClient, e.insecureSkipVerify, e.allowPrivateTargets)
-		updatedSite := site
-		updatedSite.HasSSL = result.HasSSL
-		updatedSite.CertExpiry = result.CertExpiry
-		updatedSite.Latency = time.Duration(result.LatencyNs)
-		updatedSite.LastCheck = time.Now()
-		e.handleStatusChange(updatedSite, result.Status, result.StatusCode, time.Duration(result.LatencyNs), result.ErrorReason)
-	}
-}
-
-func (e *Engine) checkPush(_ context.Context, site models.Site) {
-	if site.Status == models.StatusPending {
-		return
-	}
-
-	interval := time.Duration(site.Interval) * time.Second
-	grace := interval / 2
-	if grace < minPushGrace {
-		grace = minPushGrace
-	}
-
-	overdue := site.LastCheck.Add(interval)
-	staleMark := overdue.Add(grace / 2)
-	graceEnd := overdue.Add(grace)
-	now := time.Now()
-
-	if now.After(graceEnd) {
-		if site.Status != models.StatusDown {
-			e.handleStatusChange(site, string(models.StatusDown), 0, 0, "heartbeat missed")
-		}
-	} else if now.After(staleMark) {
-		if site.Status != models.StatusStale {
-			e.handleStatusChange(site, string(models.StatusStale), 0, 0, "heartbeat stale")
-		}
-	} else if now.After(overdue) {
-		if site.Status != models.StatusLate {
-			e.handleStatusChange(site, string(models.StatusLate), 0, 0, "heartbeat overdue")
-		}
-	}
-}
-
-func (e *Engine) checkGroup(_ context.Context, site models.Site) {
-	e.mu.RLock()
-	status := models.StatusUp
-	hasChildren := false
-	for _, child := range e.liveState {
-		if child.ParentID != site.ID || child.Type == "group" {
-			continue
-		}
-		hasChildren = true
-		if child.Paused || e.isInMaintenance(child.ID) {
-			continue
-		}
-		if child.Status == models.StatusDown || child.Status == models.StatusSSLExp {
-			status = models.StatusDown
-		} else if child.Status == models.StatusStale && status != models.StatusDown {
-			status = models.StatusStale
-		} else if child.Status == models.StatusLate && status != models.StatusDown && status != models.StatusStale {
-			status = models.StatusLate
-		} else if child.Status == models.StatusPending && status != models.StatusDown && status != models.StatusStale && status != models.StatusLate {
-			status = models.StatusPending
-		}
-	}
-	e.mu.RUnlock()
-
-	if !hasChildren {
-		status = models.StatusPending
-	}
-
-	var prev models.Status
-	e.applyState(site.ID, func(s *models.Site) {
-		prev = s.Status
-		s.Status = status
-		if status != prev && prev != models.StatusPending {
-			s.StatusChangedAt = time.Now()
-		}
-	})
-	e.recordCheck(site.ID, 0, !status.IsBroken())
-
-	if status != prev && prev != models.StatusPending {
-		e.enqueueWrite(writeStateChange{siteID: site.ID, fromStatus: string(prev), toStatus: string(status)})
-		if status.IsBroken() {
-			e.AddLog(fmt.Sprintf("Group '%s' is %s", site.Name, status))
-		} else if prev.IsBroken() {
-			e.AddLog(fmt.Sprintf("Group '%s' recovered", site.Name))
-		}
-	}
-}
-
-func (e *Engine) EnqueueProbeCheck(siteID int, nodeID string, latencyNs int64, isUp bool) {
-	e.enqueueWrite(writeProbeCheck{siteID: siteID, nodeID: nodeID, latencyNs: latencyNs, isUp: isUp})
-}
-
-// SetAggStrategy must be called before Start: the field is read by the probe
-// aggregation path without synchronization.
-func (e *Engine) SetAggStrategy(strategy AggregationStrategy) {
-	e.aggStrategy = strategy
-}
-
-func (e *Engine) IngestProbeResult(nodeID string, siteID int, latencyNs int64, isUp bool, errorReason string) {
-	e.mu.RLock()
-	site, exists := e.liveState[siteID]
-	e.mu.RUnlock()
-	if !exists {
-		return
-	}
-
-	staleAfter := time.Duration(site.Interval) * time.Second * 3
-	if staleAfter < time.Minute {
-		staleAfter = time.Minute
-	}
-
-	now := time.Now()
-	e.probeResultsMu.Lock()
-	if e.probeResults[siteID] == nil {
-		e.probeResults[siteID] = make(map[string]NodeResult)
-	}
-	e.probeResults[siteID][nodeID] = NodeResult{
-		NodeID:      nodeID,
-		IsUp:        isUp,
-		LatencyNs:   latencyNs,
-		CheckedAt:   now,
-		ErrorReason: errorReason,
-	}
-	results := make([]NodeResult, 0, len(e.probeResults[siteID]))
-	for id, r := range e.probeResults[siteID] {
-		if now.Sub(r.CheckedAt) > staleAfter {
-			delete(e.probeResults[siteID], id)
-			continue
-		}
-		results = append(results, r)
-	}
-	e.probeResultsMu.Unlock()
-
-	aggUp, avgLatency := AggregateStatus(results, e.aggStrategy)
-
-	probeStatus := models.StatusUp
-	if !aggUp {
-		probeStatus = models.StatusDown
-	}
-
-	updatedSite := site
-	updatedSite.Latency = time.Duration(avgLatency)
-	updatedSite.LastCheck = time.Now()
-	e.handleStatusChange(updatedSite, string(probeStatus), 0, time.Duration(avgLatency), errorReason)
-}
-
-func (e *Engine) GetProbeResults(siteID int) map[string]NodeResult {
-	e.probeResultsMu.RLock()
-	defer e.probeResultsMu.RUnlock()
-	src := e.probeResults[siteID]
-	cp := make(map[string]NodeResult, len(src))
-	for k, v := range src {
-		cp[k] = v
-	}
-	return cp
-}

internal/monitor/checks_test.go

@@ -1,389 +0,0 @@
-package monitor
-
-import (
-	"context"
-	"testing"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-// --- Group 2: Heartbeat ---
-
-func TestRecordHeartbeat_ValidToken(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "push-test", Type: "push", Token: "abc123"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-
-	if !e.RecordHeartbeat("abc123") {
-		t.Error("expected true for valid token")
-	}
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected UP, got %s", s.Status)
-	}
-	if time.Since(s.LastCheck) > time.Second {
-		t.Error("expected LastCheck to be recent")
-	}
-}
-
-func TestRecordHeartbeat_RecoveryFromDown(t *testing.T) {
-	ms := newMockStore()
-	ms.alerts[1] = models.AlertConfig{ID: 1, Name: "test", Type: "webhook", Settings: map[string]string{"url": "http://example.com"}}
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "push-test", Type: "push", Token: "abc123", AlertID: 1},
-		SiteState:  models.SiteState{Status: "DOWN", FailureCount: 3},
-	}
-	injectSite(e, site)
-
-	if !e.RecordHeartbeat("abc123") {
-		t.Error("expected true")
-	}
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected UP, got %s", s.Status)
-	}
-	if s.FailureCount != 0 {
-		t.Errorf("expected FailureCount 0, got %d", s.FailureCount)
-	}
-	waitAsync()
-	if len(ms.getAlertCallsSnapshot()) == 0 {
-		t.Error("expected recovery alert")
-	}
-}
-
-func TestRecordHeartbeat_UnknownToken(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-
-	if e.RecordHeartbeat("unknown") {
-		t.Error("expected false for unknown token")
-	}
-}
-
-func TestRecordHeartbeat_InactiveEngine(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Type: "push", Token: "abc123"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-	e.SetActive(false)
-
-	if e.RecordHeartbeat("abc123") {
-		t.Error("expected false when inactive")
-	}
-}
-
-// --- Group 3: Push Deadline ---
-
-func TestCheckPush_DeadlineMissed(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "push", Type: "push", Interval: 10, MaxRetries: 0},
-		SiteState:  models.SiteState{Status: "UP", LastCheck: time.Now().Add(-120 * time.Second)},
-	}
-	injectSite(e, site)
-
-	e.checkPush(context.Background(), site)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "DOWN" {
-		t.Errorf("expected DOWN after missed deadline, got %s", s.Status)
-	}
-}
-
-func TestCheckPush_OverdueBecomesLate(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "push", Type: "push", Interval: 300},
-		SiteState:  models.SiteState{Status: "UP", LastCheck: time.Now().Add(-310 * time.Second)},
-	}
-	injectSite(e, site)
-
-	e.checkPush(context.Background(), site)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "LATE" {
-		t.Errorf("expected LATE when overdue but within grace, got %s", s.Status)
-	}
-}
-
-func TestCheckPush_OverdueBecomesStale(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	// interval=300, grace=150 (300/2), staleMark=overdue+75
-	// at 380s: past staleMark(375) but before graceEnd(450)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "push", Type: "push", Interval: 300},
-		SiteState:  models.SiteState{Status: "UP", LastCheck: time.Now().Add(-380 * time.Second)},
-	}
-	injectSite(e, site)
-
-	e.checkPush(context.Background(), site)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "STALE" {
-		t.Errorf("expected STALE when past midpoint of grace, got %s", s.Status)
-	}
-}
-
-func TestCheckPush_WithinDeadline(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "push", Type: "push", Interval: 60},
-		SiteState:  models.SiteState{Status: "UP", LastCheck: time.Now()},
-	}
-	injectSite(e, site)
-
-	e.checkPush(context.Background(), site)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected UP, got %s", s.Status)
-	}
-}
-
-func TestCheckPush_PendingStaysPending(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "push", Type: "push", Interval: 60},
-		SiteState:  models.SiteState{Status: "PENDING"},
-	}
-	injectSite(e, site)
-
-	e.checkPush(context.Background(), site)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "PENDING" {
-		t.Errorf("expected PENDING to stay until first heartbeat, got %s", s.Status)
-	}
-}
-
-// --- Group 4: Group Checks ---
-
-func TestCheckGroup_AllChildrenUp(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	group := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "group", Type: "group"},
-		SiteState:  models.SiteState{Status: "PENDING"},
-	}
-	child1 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 2, Name: "child1", Type: "http", ParentID: 1},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	child2 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 3, Name: "child2", Type: "http", ParentID: 1},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, group)
-	injectSite(e, child1)
-	injectSite(e, child2)
-
-	e.checkGroup(context.Background(), group)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected group UP, got %s", s.Status)
-	}
-}
-
-func TestCheckGroup_OneChildDown(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	group := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "group", Type: "group"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	child1 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 2, Name: "child1", Type: "http", ParentID: 1},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	child2 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 3, Name: "child2", Type: "http", ParentID: 1},
-		SiteState:  models.SiteState{Status: "DOWN"},
-	}
-	injectSite(e, group)
-	injectSite(e, child1)
-	injectSite(e, child2)
-
-	e.checkGroup(context.Background(), group)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "DOWN" {
-		t.Errorf("expected group DOWN, got %s", s.Status)
-	}
-}
-
-func TestCheckGroup_PausedChildIgnored(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	group := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "group", Type: "group"},
-	}
-	child1 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 2, Name: "child1", Type: "http", ParentID: 1},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	child2 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 3, Name: "child2", Type: "http", ParentID: 1, Paused: true},
-		SiteState:  models.SiteState{Status: "DOWN"},
-	}
-	injectSite(e, group)
-	injectSite(e, child1)
-	injectSite(e, child2)
-
-	e.checkGroup(context.Background(), group)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected UP (paused child ignored), got %s", s.Status)
-	}
-}
-
-func TestCheckGroup_MaintenanceChildIgnored(t *testing.T) {
-	ms := newMockStore()
-	ms.maintenance[3] = true
-	e := newTestEngine(ms)
-	group := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "group", Type: "group"},
-	}
-	child1 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 2, Name: "child1", Type: "http", ParentID: 1},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	child2 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 3, Name: "child2", Type: "http", ParentID: 1},
-		SiteState:  models.SiteState{Status: "DOWN"},
-	}
-	injectSite(e, group)
-	injectSite(e, child1)
-	injectSite(e, child2)
-	e.refreshMaintenanceCache(context.Background())
-
-	e.checkGroup(context.Background(), group)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "UP" {
-		t.Errorf("expected UP (maint child ignored), got %s", s.Status)
-	}
-}
-
-func TestCheckGroup_NoChildren(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	group := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "group", Type: "group"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, group)
-
-	e.checkGroup(context.Background(), group)
-
-	s, _ := getSite(e, 1)
-	if s.Status != "PENDING" {
-		t.Errorf("expected PENDING for no children, got %s", s.Status)
-	}
-}
-
-// Groups must not auto-pause when all children are paused — that creates a
-// one-way trap because monitorRoutine skips paused sites.
-func TestCheckGroup_AllPausedNoAutoFreeze(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	group := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "group", Type: "group"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	child1 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 2, Name: "child1", Type: "http", ParentID: 1, Paused: true},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	child2 := models.Site{
-		SiteConfig: models.SiteConfig{ID: 3, Name: "child2", Type: "http", ParentID: 1, Paused: true},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, group)
-	injectSite(e, child1)
-	injectSite(e, child2)
-
-	e.checkGroup(context.Background(), group)
-
-	s, _ := getSite(e, 1)
-	if s.Paused {
-		t.Error("group must not auto-pause when all children are paused")
-	}
-}
-
-// Dead probe results must be expired so they don't poison aggregation.
-func TestIngestProbeResult_ExpiresStaleProbes(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", Type: "http", Interval: 30},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-
-	e.probeResultsMu.Lock()
-	e.probeResults[1] = map[string]NodeResult{
-		"dead-probe": {
-			NodeID:    "dead-probe",
-			IsUp:      false,
-			CheckedAt: time.Now().Add(-10 * time.Minute),
-		},
-	}
-	e.probeResultsMu.Unlock()
-
-	e.IngestProbeResult("live-probe", 1, 5000, true, "")
-
-	e.probeResultsMu.RLock()
-	_, deadExists := e.probeResults[1]["dead-probe"]
-	_, liveExists := e.probeResults[1]["live-probe"]
-	e.probeResultsMu.RUnlock()
-
-	if deadExists {
-		t.Error("stale probe result should have been expired")
-	}
-	if !liveExists {
-		t.Error("live probe result should still exist")
-	}
-}
-
-// RemoveSite must clean up probeResults.
-func TestRemoveSite_CleansProbeResults(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", Type: "http"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-
-	e.probeResultsMu.Lock()
-	e.probeResults[1] = map[string]NodeResult{
-		"node-a": {NodeID: "node-a", IsUp: true, CheckedAt: time.Now()},
-	}
-	e.probeResultsMu.Unlock()
-
-	e.RemoveSite(1)
-
-	e.probeResultsMu.RLock()
-	defer e.probeResultsMu.RUnlock()
-	if _, exists := e.probeResults[1]; exists {
-		t.Error("probe results should be cleaned up after RemoveSite")
-	}
-}

internal/monitor/history.go

@@ -14,8 +14,8 @@ type SiteHistory struct {
 	UpChecks    int
 }
 
-func (e *Engine) InitHistory(ctx context.Context) {
-	all, err := e.db.LoadAllHistory(ctx, maxHistoryLen)
+func (e *Engine) InitHistory() {
+	all, err := e.db.LoadAllHistory(context.Background(), maxHistoryLen)
 	if err != nil {
 		e.AddLog("Failed to load check history: " + err.Error())
 		return

internal/monitor/maintenance.go

@@ -1,70 +0,0 @@
-package monitor
-
-import (
-	"context"
-	"fmt"
-	"time"
-)
-
-func (e *Engine) maintenancePruner(ctx context.Context) {
-	ticker := time.NewTicker(maintPruneInterval)
-	defer ticker.Stop()
-
-	e.pruneMaintenanceWindows(ctx)
-
-	for {
-		select {
-		case <-ticker.C:
-			e.pruneMaintenanceWindows(ctx)
-		case <-ctx.Done():
-			return
-		}
-	}
-}
-
-func (e *Engine) pruneMaintenanceWindows(ctx context.Context) {
-	pruned, err := e.db.PruneExpiredMaintenanceWindows(ctx, e.maintRetention)
-	if err != nil {
-		e.AddLog(fmt.Sprintf("Maintenance prune error: %v", err))
-		return
-	}
-	if pruned > 0 {
-		e.AddLog(fmt.Sprintf("Pruned %d expired maintenance window(s)", pruned))
-	}
-}
-
-func (e *Engine) isInMaintenance(monitorID int) bool {
-	e.maintCacheMu.RLock()
-	defer e.maintCacheMu.RUnlock()
-	return e.maintCache[monitorID]
-}
-
-func (e *Engine) refreshMaintenanceCache(ctx context.Context) {
-	windows, err := e.db.GetActiveMaintenanceWindows(ctx)
-	if err != nil {
-		return
-	}
-
-	directMaint := make(map[int]bool)
-	var globalMaint bool
-	for _, w := range windows {
-		if w.MonitorID == 0 {
-			globalMaint = true
-		} else {
-			directMaint[w.MonitorID] = true
-		}
-	}
-
-	resolved := make(map[int]bool)
-	e.mu.RLock()
-	for id, site := range e.liveState {
-		if globalMaint || directMaint[id] || (site.ParentID > 0 && directMaint[site.ParentID]) {
-			resolved[id] = true
-		}
-	}
-	e.mu.RUnlock()
-
-	e.maintCacheMu.Lock()
-	e.maintCache = resolved
-	e.maintCacheMu.Unlock()
-}

internal/monitor/maintenance_test.go

@@ -1,53 +0,0 @@
-package monitor
-
-import (
-	"context"
-	"testing"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-// Maintenance cache resolves parent relationships correctly.
-func TestIsInMaintenance_UsesCache(t *testing.T) {
-	ms := newMockStore()
-	ms.maintenance[10] = true // direct maintenance on group
-	e := newTestEngine(ms)
-	group := models.Site{
-		SiteConfig: models.SiteConfig{ID: 10, Name: "group", Type: "group"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	child := models.Site{
-		SiteConfig: models.SiteConfig{ID: 20, Name: "child", Type: "http", ParentID: 10},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, group)
-	injectSite(e, child)
-	e.refreshMaintenanceCache(context.Background())
-
-	if !e.isInMaintenance(10) {
-		t.Error("group should be in maintenance (direct)")
-	}
-	if !e.isInMaintenance(20) {
-		t.Error("child should be in maintenance (parent)")
-	}
-	if e.isInMaintenance(99) {
-		t.Error("unknown monitor should not be in maintenance")
-	}
-}
-
-// Global maintenance (monitor_id=0) applies to all monitors.
-func TestIsInMaintenance_GlobalMaintenance(t *testing.T) {
-	ms := newMockStore()
-	ms.maintenance[0] = true
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", Type: "http"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-	e.refreshMaintenanceCache(context.Background())
-
-	if !e.isInMaintenance(1) {
-		t.Error("all monitors should be in maintenance during global window")
-	}
-}

internal/monitor/monitor.go

@@ -4,12 +4,14 @@ import (
 	"context"
 	"crypto/tls"
 	"fmt"
+	"math/rand/v2"
 	"net/http"
 	"regexp"
 	"strings"
 	"sync"
 	"time"
 
+	"gitea.lerkolabs.com/lerkolabs/uptop/internal/alert"
 	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
 	"gitea.lerkolabs.com/lerkolabs/uptop/internal/store"
 )
@@ -22,16 +24,23 @@ const (
 	maintPruneInterval    = 15 * time.Minute
 	defaultMaintRetention = 7 * 24 * time.Hour
 	dbWriteBuffer         = 4096
-	alertSendTimeout      = 30 * time.Second
 	dbPruneInterval       = 10 * time.Minute
 )
 
+type AlertHealth struct {
+	LastSendAt time.Time
+	LastSendOK bool
+	LastError  string
+	SendCount  int
+	FailCount  int
+}
+
 type Engine struct {
 	mu        sync.RWMutex
 	liveState map[int]models.Site
 
 	logMu    sync.RWMutex
-	logStore []models.LogEntry
+	logStore []string
 
 	activeMu sync.RWMutex
 	isActive bool
@@ -64,7 +73,6 @@ type Engine struct {
 	dbWrites  chan dbWrite
 	writerWG  sync.WaitGroup
 	checkerWG sync.WaitGroup
-	ctx       context.Context
 	cancel    context.CancelFunc
 	stopOnce  sync.Once
 }
@@ -91,7 +99,6 @@ func newEngine(s store.Store, allowPrivateTargets bool) *Engine {
 		allowPrivateTargets: allowPrivateTargets,
 		maintRetention:      defaultMaintRetention,
 		dbWrites:            make(chan dbWrite, dbWriteBuffer),
-		ctx:                 context.Background(),
 		db:                  s,
 		strictClient: &http.Client{
 			Transport: &http.Transport{
@@ -145,13 +152,11 @@ func fmtDurationShort(d time.Duration) string {
 // appendLog adds a timestamped entry to the in-memory ring buffer and returns
 // it. It never touches the database, so it is safe to call from the db-write
 // drop/error path without recursing back through the write queue.
-func (e *Engine) appendLog(msg string) models.LogEntry {
-	entry := models.LogEntry{
-		Message:   sanitizeLog(msg),
-		CreatedAt: time.Now(),
-	}
+func (e *Engine) appendLog(msg string) string {
+	ts := time.Now().Format("15:04:05")
+	entry := fmt.Sprintf("[%s] %s", ts, sanitizeLog(msg))
 	e.logMu.Lock()
-	e.logStore = append([]models.LogEntry{entry}, e.logStore...)
+	e.logStore = append([]string{entry}, e.logStore...)
 	if len(e.logStore) > maxLogEntries {
 		e.logStore = e.logStore[:maxLogEntries]
 	}
@@ -161,7 +166,7 @@ func (e *Engine) appendLog(msg string) models.LogEntry {
 
 func (e *Engine) AddLog(msg string) {
 	entry := e.appendLog(msg)
-	e.enqueueWrite(writeLog{message: entry.Message})
+	e.enqueueWrite(writeLog{message: entry})
 }
 
 // enqueueWrite hands a persistence task to the writer goroutine without
@@ -202,8 +207,6 @@ func (e *Engine) dbWriter(ctx context.Context) {
 }
 
 // drainWrites flushes everything still buffered, best-effort, at shutdown.
-// Uses context.Background because the engine ctx is already cancelled when
-// this runs — writes still need to reach the DB.
 func (e *Engine) drainWrites() {
 	for {
 		select {
@@ -242,33 +245,169 @@ func (e *Engine) Stop() {
 	})
 }
 
-func (e *Engine) InitLogs(ctx context.Context) {
-	entries, err := e.db.LoadLogs(ctx, maxLogEntries)
+func (e *Engine) InitLogs() {
+	logs, err := e.db.LoadLogs(context.Background(), maxLogEntries)
 	if err != nil {
 		return
 	}
-	if len(entries) == 0 {
+	if len(logs) == 0 {
 		return
 	}
 	e.logMu.Lock()
 	defer e.logMu.Unlock()
-	e.logStore = entries
+	e.logStore = logs
 }
 
-func (e *Engine) GetLogs() []models.LogEntry {
+// InitAlertHealth restores persisted alert send health so the dashboard shows real
+// "last sent" / health state on startup instead of resetting every channel to "never".
+func (e *Engine) InitAlertHealth() {
+	records, err := e.db.LoadAlertHealth(context.Background())
+	if err != nil {
+		return
+	}
+	e.alertHealthMu.Lock()
+	defer e.alertHealthMu.Unlock()
+	for id, r := range records {
+		e.alertHealth[id] = AlertHealth{
+			LastSendAt: r.LastSendAt,
+			LastSendOK: r.LastSendOK,
+			LastError:  r.LastError,
+			SendCount:  r.SendCount,
+			FailCount:  r.FailCount,
+		}
+	}
+}
+
+func (e *Engine) GetLogs() []string {
 	e.logMu.RLock()
 	defer e.logMu.RUnlock()
-	logs := make([]models.LogEntry, len(e.logStore))
+	logs := make([]string, len(e.logStore))
 	copy(logs, e.logStore)
 	return logs
 }
 
+func (e *Engine) SetActive(active bool) {
+	e.activeMu.Lock()
+	defer e.activeMu.Unlock()
+	if e.isActive != active {
+		e.isActive = active
+		status := "RESUMED (Active)"
+		if !active {
+			status = "PAUSED (Passive)"
+		}
+		e.AddLog(fmt.Sprintf("Engine %s", status))
+	}
+}
+
+func (e *Engine) IsActive() bool {
+	e.activeMu.RLock()
+	defer e.activeMu.RUnlock()
+	return e.isActive
+}
+
+func (e *Engine) GetAllSites() []models.Site {
+	e.mu.RLock()
+	defer e.mu.RUnlock()
+	sites := make([]models.Site, 0, len(e.liveState))
+	for _, s := range e.liveState {
+		sites = append(sites, s)
+	}
+	return sites
+}
+
+func (e *Engine) GetLiveState() map[int]models.Site {
+	e.mu.RLock()
+	defer e.mu.RUnlock()
+	cp := make(map[int]models.Site, len(e.liveState))
+	for k, v := range e.liveState {
+		cp[k] = v
+	}
+	return cp
+}
+
+func (e *Engine) RecordHeartbeat(token string) bool {
+	if !e.IsActive() {
+		return false
+	}
+
+	e.mu.RLock()
+	targetID, ok := e.tokenIndex[token]
+	e.mu.RUnlock()
+	if !ok {
+		return false
+	}
+
+	var (
+		prevStatus models.Status
+		name       string
+		alertID    int
+		downSince  time.Time
+	)
+	_, exists := e.applyState(targetID, func(s *models.Site) {
+		prevStatus = s.Status
+		name = s.Name
+		alertID = s.AlertID
+		downSince = s.StatusChangedAt // captured before mutation = when it went down
+
+		s.LastCheck = time.Now()
+		s.Status = models.StatusUp
+		s.FailureCount = 0
+		s.Latency = 0
+		s.LastError = ""
+		s.LastSuccessAt = time.Now()
+		if prevStatus != models.StatusUp {
+			s.StatusChangedAt = time.Now()
+		}
+	})
+	if !exists {
+		return false
+	}
+
+	switch prevStatus {
+	case models.StatusPending:
+		e.AddLog(fmt.Sprintf("Push Monitor '%s' received first heartbeat", name))
+	case models.StatusLate:
+		e.AddLog(fmt.Sprintf("Push Monitor '%s' heartbeat arrived (was late)", name))
+	case models.StatusStale:
+		e.AddLog(fmt.Sprintf("Push Monitor '%s' heartbeat arrived (was stale)", name))
+	case models.StatusDown:
+		downDur := ""
+		if !downSince.IsZero() {
+			downDur = fmt.Sprintf(" (was down %s)", fmtDurationShort(time.Since(downSince)))
+		}
+		e.AddLog(fmt.Sprintf("Push Monitor '%s' recovered%s", name, downDur))
+		go e.triggerAlert(alertID, "✅ RECOVERY", fmt.Sprintf("Push Monitor '%s' is receiving heartbeats.%s", name, downDur))
+	}
+
+	e.recordCheck(targetID, 0, true)
+
+	if prevStatus != models.StatusUp && prevStatus != models.StatusPending {
+		e.enqueueWrite(writeStateChange{siteID: targetID, fromStatus: string(prevStatus), toStatus: string(models.StatusUp)})
+	}
+
+	return true
+}
+
+func (e *Engine) addToTokenIndex(site models.Site) {
+	if site.Type == "push" && site.Token != "" {
+		e.tokenIndex[site.Token] = site.ID
+	}
+}
+
+func (e *Engine) removeFromTokenIndex(id int) {
+	for token, sid := range e.tokenIndex {
+		if sid == id {
+			delete(e.tokenIndex, token)
+			return
+		}
+	}
+}
+
 func (e *Engine) Start(ctx context.Context) {
 	// e.cancel is invoked by Stop() to drain and halt the writer; gosec can't
 	// trace the cross-method call, and cancelling the parent reaps this child
 	// regardless, so the leak it warns about can't occur.
 	ctx, e.cancel = context.WithCancel(ctx) //nolint:gosec // cancel is called in Stop()
-	e.ctx = ctx
 
 	e.writerWG.Add(1)
 	go e.dbWriter(ctx)
@@ -283,6 +422,8 @@ func (e *Engine) Start(ctx context.Context) {
 			default:
 			}
 
+			e.refreshMaintenanceCache(ctx)
+
 			configs, err := e.db.GetSites(ctx)
 			if err != nil {
 				e.AddLog(fmt.Sprintf("Failed to load sites: %v", err))
@@ -325,10 +466,6 @@ func (e *Engine) Start(ctx context.Context) {
 				}
 			}
 
-			// Refresh after sites load so the cache covers newly added sites.
-			// On first iteration liveState was empty before the loop above.
-			e.refreshMaintenanceCache(ctx)
-
 			e.mu.RLock()
 			var vanished []int
 			for id := range e.liveState {
@@ -356,3 +493,637 @@ func (e *Engine) Start(ctx context.Context) {
 		e.maintenancePruner(ctx)
 	}()
 }
+
+func (e *Engine) maintenancePruner(ctx context.Context) {
+	ticker := time.NewTicker(maintPruneInterval)
+	defer ticker.Stop()
+
+	e.pruneMaintenanceWindows(ctx)
+
+	for {
+		select {
+		case <-ticker.C:
+			e.pruneMaintenanceWindows(ctx)
+		case <-ctx.Done():
+			return
+		}
+	}
+}
+
+func (e *Engine) pruneMaintenanceWindows(ctx context.Context) {
+	pruned, err := e.db.PruneExpiredMaintenanceWindows(ctx, e.maintRetention)
+	if err != nil {
+		e.AddLog(fmt.Sprintf("Maintenance prune error: %v", err))
+		return
+	}
+	if pruned > 0 {
+		e.AddLog(fmt.Sprintf("Pruned %d expired maintenance window(s)", pruned))
+	}
+}
+
+func (e *Engine) UpdateSiteConfig(cfg models.SiteConfig) {
+	e.mu.Lock()
+	if existing, ok := e.liveState[cfg.ID]; ok {
+		e.removeFromTokenIndex(cfg.ID)
+		existing.SiteConfig = cfg
+		e.liveState[cfg.ID] = existing
+		e.addToTokenIndex(existing)
+	}
+	e.mu.Unlock()
+
+	e.signalRecheck(cfg.ID)
+}
+
+func (e *Engine) getRecheckChan(id int) chan struct{} {
+	e.recheckMu.Lock()
+	defer e.recheckMu.Unlock()
+	ch, ok := e.recheck[id]
+	if !ok {
+		ch = make(chan struct{}, 1)
+		e.recheck[id] = ch
+	}
+	return ch
+}
+
+func (e *Engine) signalRecheck(id int) {
+	ch := e.getRecheckChan(id)
+	select {
+	case ch <- struct{}{}:
+	default:
+	}
+}
+
+func (e *Engine) RemoveSite(id int) {
+	e.mu.Lock()
+	e.removeFromTokenIndex(id)
+	delete(e.liveState, id)
+	e.mu.Unlock()
+	e.removeHistory(id)
+
+	e.probeResultsMu.Lock()
+	delete(e.probeResults, id)
+	e.probeResultsMu.Unlock()
+
+	e.recheckMu.Lock()
+	delete(e.recheck, id)
+	e.recheckMu.Unlock()
+}
+
+func (e *Engine) ToggleSitePause(id int) bool {
+	var (
+		paused bool
+		name   string
+	)
+	_, ok := e.applyState(id, func(s *models.Site) {
+		s.Paused = !s.Paused
+		paused = s.Paused
+		name = s.Name
+	})
+	if !ok {
+		return false
+	}
+	if paused {
+		e.AddLog(fmt.Sprintf("Monitor '%s' paused", name))
+	} else {
+		e.AddLog(fmt.Sprintf("Monitor '%s' resumed", name))
+	}
+	return paused
+}
+
+func (e *Engine) monitorRoutine(ctx context.Context, id int) {
+	recheckCh := e.getRecheckChan(id)
+
+	// Stagger initial check to avoid thundering herd on startup
+	stagger := time.Duration(rand.IntN(3000)) * time.Millisecond //nolint:gosec // non-security jitter
+	select {
+	case <-time.After(stagger):
+	case <-ctx.Done():
+		return
+	}
+
+	e.checkByID(ctx, id)
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		default:
+		}
+
+		if !e.IsActive() {
+			select {
+			case <-time.After(pollInterval):
+			case <-ctx.Done():
+				return
+			case <-recheckCh:
+			}
+			continue
+		}
+
+		e.mu.RLock()
+		site, exists := e.liveState[id]
+		e.mu.RUnlock()
+		if !exists {
+			return
+		}
+
+		if site.Paused {
+			select {
+			case <-time.After(pollInterval):
+			case <-ctx.Done():
+				return
+			case <-recheckCh:
+			}
+			continue
+		}
+
+		interval := site.Interval
+		if interval < minCheckInterval {
+			interval = minCheckInterval
+		}
+		jitter := time.Duration(rand.IntN(interval*100)) * time.Millisecond //nolint:gosec // non-security jitter
+		select {
+		case <-time.After(time.Duration(interval)*time.Second + jitter):
+		case <-ctx.Done():
+			return
+		case <-recheckCh:
+		}
+		e.checkByID(ctx, id)
+	}
+}
+
+// applyState atomically reads, mutates, and writes back the live entry for id.
+// The mutator runs under the engine write lock and receives a pointer to the
+// CURRENT live state, so concurrent config edits, pauses, and heartbeats are
+// never clobbered by a stale snapshot. The mutator must only touch runtime /
+// check-result fields — config fields (Name/URL/Type/Token/Interval/AlertID/…)
+// are owned by UpdateSiteConfig and must not be written here. Returns the
+// post-mutation copy and whether the site still exists.
+func (e *Engine) applyState(id int, mutate func(s *models.Site)) (models.Site, bool) {
+	e.mu.Lock()
+	defer e.mu.Unlock()
+	cur, ok := e.liveState[id]
+	if !ok {
+		return models.Site{}, false
+	}
+	mutate(&cur)
+	e.liveState[id] = cur
+	return cur, true
+}
+
+func (e *Engine) checkByID(ctx context.Context, id int) {
+	if !e.IsActive() {
+		return
+	}
+
+	e.mu.RLock()
+	site, exists := e.liveState[id]
+	e.mu.RUnlock()
+	if !exists || site.Paused {
+		return
+	}
+
+	switch site.Type {
+	case "push":
+		e.checkPush(ctx, site)
+	case "group":
+		e.checkGroup(ctx, site)
+	default:
+		result := RunCheck(ctx, site.SiteConfig, e.strictClient, e.insecureClient, e.insecureSkipVerify, e.allowPrivateTargets)
+		updatedSite := site
+		updatedSite.HasSSL = result.HasSSL
+		updatedSite.CertExpiry = result.CertExpiry
+		updatedSite.Latency = time.Duration(result.LatencyNs)
+		updatedSite.LastCheck = time.Now()
+		e.handleStatusChange(updatedSite, result.Status, result.StatusCode, time.Duration(result.LatencyNs), result.ErrorReason)
+	}
+}
+
+func (e *Engine) checkPush(_ context.Context, site models.Site) {
+	if site.Status == models.StatusPending {
+		return
+	}
+
+	interval := time.Duration(site.Interval) * time.Second
+	grace := interval / 2
+	if grace < minPushGrace {
+		grace = minPushGrace
+	}
+
+	overdue := site.LastCheck.Add(interval)
+	staleMark := overdue.Add(grace / 2)
+	graceEnd := overdue.Add(grace)
+	now := time.Now()
+
+	if now.After(graceEnd) {
+		if site.Status != models.StatusDown {
+			e.handleStatusChange(site, string(models.StatusDown), 0, 0, "heartbeat missed")
+		}
+	} else if now.After(staleMark) {
+		if site.Status != models.StatusStale {
+			e.handleStatusChange(site, string(models.StatusStale), 0, 0, "heartbeat stale")
+		}
+	} else if now.After(overdue) {
+		if site.Status != models.StatusLate {
+			e.handleStatusChange(site, string(models.StatusLate), 0, 0, "heartbeat overdue")
+		}
+	}
+}
+
+// handleStatusChange folds a check result into the live state. snap is the
+// stale snapshot the check ran against; the actual mutation is applied onto the
+// CURRENT live entry via applyState, so a concurrent pause / config edit /
+// heartbeat is never reverted by this write. Logs and alerts are emitted after
+// the lock is released, off the critical section.
+func (e *Engine) handleStatusChange(snap models.Site, rawStatus string, code int, latency time.Duration, errorReason string) {
+	if !e.IsActive() {
+		return
+	}
+
+	inMaint := e.isInMaintenance(snap.ID)
+	status := models.Status(rawStatus)
+
+	var (
+		prev, next            models.Status
+		name, typ             string
+		alertID               int
+		failCount, maxRetries int
+		confirmedDown         bool
+		failedCheck           bool
+		downSince             time.Time
+		sslWarnFire           bool
+		sslDays               int
+		skipped               bool
+		changed               bool
+	)
+
+	_, exists := e.applyState(snap.ID, func(s *models.Site) {
+		// A non-UP result computed from a stale snapshot must not override a
+		// heartbeat (or newer check) that landed while we were evaluating.
+		if status != models.StatusUp && s.LastCheck.After(snap.LastCheck) {
+			skipped = true
+			return
+		}
+
+		prev = s.Status
+		name = s.Name
+		typ = s.Type
+		alertID = s.AlertID
+		maxRetries = s.MaxRetries
+		downSince = s.StatusChangedAt
+
+		// Fresh check results (measured by the run against snap).
+		s.StatusCode = code
+		s.Latency = snap.Latency
+		s.LastCheck = snap.LastCheck
+		s.HasSSL = snap.HasSSL
+		s.CertExpiry = snap.CertExpiry
+		s.LastError = errorReason
+		if status == models.StatusUp {
+			s.LastSuccessAt = time.Now()
+			s.LastError = ""
+		}
+
+		// Status + failure-count transition, based on the CURRENT live status.
+		if status == models.StatusUp {
+			s.FailureCount = 0
+			s.Status = models.StatusUp
+		} else {
+			if s.FailureCount <= s.MaxRetries {
+				s.FailureCount++
+			}
+			if s.FailureCount > s.MaxRetries {
+				if s.Status != status {
+					confirmedDown = true
+				}
+				s.Status = status
+				s.FailureCount = s.MaxRetries + 1
+			} else {
+				failedCheck = true
+			}
+		}
+		failCount = s.FailureCount
+
+		if s.Status != prev && prev != models.StatusPending {
+			s.StatusChangedAt = time.Now()
+		} else if s.StatusChangedAt.IsZero() && s.Status != models.StatusPending {
+			s.StatusChangedAt = time.Now()
+		}
+
+		// SSL expiry warning (fresh HasSSL/CertExpiry + config threshold).
+		if typ == "http" && s.CheckSSL && s.HasSSL {
+			days := int(time.Until(s.CertExpiry).Hours() / 24)
+			if days <= s.ExpiryThreshold && !s.SentSSLWarning && status != models.StatusSSLExp {
+				sslWarnFire = true
+				sslDays = days
+				s.SentSSLWarning = true
+			} else if days > s.ExpiryThreshold {
+				s.SentSSLWarning = false
+			}
+		}
+
+		next = s.Status
+		changed = next != prev
+	})
+
+	if !exists || skipped {
+		return
+	}
+
+	e.recordCheck(snap.ID, latency, status == models.StatusUp)
+
+	if confirmedDown {
+		if errorReason != "" {
+			e.AddLog(fmt.Sprintf("Monitor '%s' confirmed DOWN: %s", name, errorReason))
+		} else {
+			e.AddLog(fmt.Sprintf("Monitor '%s' confirmed DOWN", name))
+		}
+	} else if failedCheck {
+		e.AddLog(fmt.Sprintf("Monitor '%s' failed check %d/%d", name, failCount, maxRetries))
+	}
+
+	if changed && prev != models.StatusPending {
+		e.enqueueWrite(writeStateChange{siteID: snap.ID, fromStatus: string(prev), toStatus: string(next), reason: errorReason})
+	}
+
+	if sslWarnFire {
+		if !inMaint {
+			e.triggerAlert(alertID, "SSL WARNING", fmt.Sprintf("SSL for '%s' expires in %d days", name, sslDays))
+		} else {
+			e.AddLog(fmt.Sprintf("SSL warning for '%s' suppressed (maintenance)", name))
+		}
+	}
+
+	if prev == models.StatusUp && next == models.StatusLate {
+		e.AddLog(fmt.Sprintf("Monitor '%s' heartbeat overdue", name))
+	}
+
+	if !prev.IsBroken() && next.IsBroken() && next != models.StatusPending {
+		if inMaint {
+			e.AddLog(fmt.Sprintf("Monitor '%s' is DOWN (alerts suppressed — maintenance)", name))
+		} else {
+			msg := fmt.Sprintf("Monitor '%s' is DOWN (%s)", name, rawStatus)
+			if errorReason != "" {
+				msg = fmt.Sprintf("Monitor '%s' is DOWN: %s", name, errorReason)
+			}
+			if typ == "push" {
+				msg = fmt.Sprintf("Push Monitor '%s' missed heartbeat.", name)
+			}
+			e.triggerAlert(alertID, "🚨 ALERT", msg)
+		}
+	}
+	if prev.IsBroken() && next == models.StatusUp {
+		downDur := ""
+		if !downSince.IsZero() {
+			downDur = fmt.Sprintf(" (was down %s)", fmtDurationShort(time.Since(downSince)))
+		}
+		e.AddLog(fmt.Sprintf("Monitor '%s' recovered%s", name, downDur))
+		if !inMaint {
+			e.triggerAlert(alertID, "✅ RECOVERY", fmt.Sprintf("Monitor '%s' is UP%s", name, downDur))
+		}
+	}
+	if prev == models.StatusLate && next == models.StatusUp && !prev.IsBroken() {
+		e.AddLog(fmt.Sprintf("Monitor '%s' heartbeat arrived (was late)", name))
+	}
+}
+
+func (e *Engine) triggerAlert(alertID int, title, message string) {
+	if alertID <= 0 {
+		return
+	}
+	cfg, err := e.db.GetAlert(context.Background(), alertID)
+	if err != nil {
+		e.AddLog(fmt.Sprintf("Failed to load alert config %d: %v", alertID, err))
+		return
+	}
+	provider := alert.GetProvider(cfg)
+	if provider != nil {
+		go func() {
+			ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+			defer cancel()
+			if err := provider.Send(ctx, title, message); err != nil {
+				e.AddLog(fmt.Sprintf("Alert send failed (%s): %v", cfg.Name, err))
+				e.recordAlertResult(alertID, false, err.Error())
+			} else {
+				e.recordAlertResult(alertID, true, "")
+			}
+		}()
+	}
+}
+
+func (e *Engine) recordAlertResult(alertID int, ok bool, errMsg string) {
+	e.alertHealthMu.Lock()
+	defer e.alertHealthMu.Unlock()
+	h := e.alertHealth[alertID]
+	h.LastSendAt = time.Now()
+	h.LastSendOK = ok
+	h.SendCount++
+	if ok {
+		h.LastError = ""
+	} else {
+		h.LastError = errMsg
+		h.FailCount++
+	}
+	e.alertHealth[alertID] = h
+
+	// Persist so health survives restarts; DB IO off the alert path.
+	e.enqueueWrite(writeAlertHealth{rec: models.AlertHealthRecord{
+		AlertID:    alertID,
+		LastSendAt: h.LastSendAt,
+		LastSendOK: h.LastSendOK,
+		LastError:  h.LastError,
+		SendCount:  h.SendCount,
+		FailCount:  h.FailCount,
+	}})
+}
+
+func (e *Engine) GetAlertHealth(alertID int) AlertHealth {
+	e.alertHealthMu.RLock()
+	defer e.alertHealthMu.RUnlock()
+	return e.alertHealth[alertID]
+}
+
+func (e *Engine) TestAlert(alertID int) error {
+	cfg, err := e.db.GetAlert(context.Background(), alertID)
+	if err != nil {
+		return fmt.Errorf("failed to load alert: %w", err)
+	}
+	provider := alert.GetProvider(cfg)
+	if provider == nil {
+		return fmt.Errorf("no provider for type %q", cfg.Type)
+	}
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer cancel()
+	err = provider.Send(ctx, "🧪 Test Alert", fmt.Sprintf("Test notification from uptop for channel '%s'.", cfg.Name))
+	if err != nil {
+		e.recordAlertResult(alertID, false, err.Error())
+		return err
+	}
+	e.recordAlertResult(alertID, true, "")
+	e.AddLog(fmt.Sprintf("Test alert sent to '%s'", cfg.Name))
+	return nil
+}
+
+func (e *Engine) isInMaintenance(monitorID int) bool {
+	e.maintCacheMu.RLock()
+	defer e.maintCacheMu.RUnlock()
+	return e.maintCache[monitorID]
+}
+
+func (e *Engine) refreshMaintenanceCache(ctx context.Context) {
+	windows, err := e.db.GetActiveMaintenanceWindows(ctx)
+	if err != nil {
+		return
+	}
+
+	directMaint := make(map[int]bool)
+	var globalMaint bool
+	for _, w := range windows {
+		if w.MonitorID == 0 {
+			globalMaint = true
+		} else {
+			directMaint[w.MonitorID] = true
+		}
+	}
+
+	resolved := make(map[int]bool)
+	e.mu.RLock()
+	for id, site := range e.liveState {
+		if globalMaint || directMaint[id] || (site.ParentID > 0 && directMaint[site.ParentID]) {
+			resolved[id] = true
+		}
+	}
+	e.mu.RUnlock()
+
+	e.maintCacheMu.Lock()
+	e.maintCache = resolved
+	e.maintCacheMu.Unlock()
+}
+
+func (e *Engine) GetDisplayStatus(site models.Site) string {
+	if site.Paused {
+		return "PAUSED"
+	}
+	if e.isInMaintenance(site.ID) {
+		return "MAINT"
+	}
+	return string(site.Status)
+}
+
+func (e *Engine) checkGroup(_ context.Context, site models.Site) {
+	e.mu.RLock()
+	status := models.StatusUp
+	hasChildren := false
+	for _, child := range e.liveState {
+		if child.ParentID != site.ID || child.Type == "group" {
+			continue
+		}
+		hasChildren = true
+		if child.Paused || e.isInMaintenance(child.ID) {
+			continue
+		}
+		if child.Status == models.StatusDown || child.Status == models.StatusSSLExp {
+			status = models.StatusDown
+		} else if child.Status == models.StatusStale && status != models.StatusDown {
+			status = models.StatusStale
+		} else if child.Status == models.StatusLate && status != models.StatusDown && status != models.StatusStale {
+			status = models.StatusLate
+		} else if child.Status == models.StatusPending && status != models.StatusDown && status != models.StatusStale && status != models.StatusLate {
+			status = models.StatusPending
+		}
+	}
+	e.mu.RUnlock()
+
+	if !hasChildren {
+		status = models.StatusPending
+	}
+
+	e.applyState(site.ID, func(s *models.Site) {
+		s.Status = status
+	})
+	e.recordCheck(site.ID, 0, !status.IsBroken())
+}
+
+func (e *Engine) EnqueueProbeCheck(siteID int, nodeID string, latencyNs int64, isUp bool) {
+	e.enqueueWrite(writeProbeCheck{siteID: siteID, nodeID: nodeID, latencyNs: latencyNs, isUp: isUp})
+}
+
+// SetAggStrategy must be called before Start: the field is read by the probe
+// aggregation path without synchronization.
+func (e *Engine) SetAggStrategy(strategy AggregationStrategy) {
+	e.aggStrategy = strategy
+}
+
+func (e *Engine) IngestProbeResult(nodeID string, siteID int, latencyNs int64, isUp bool, errorReason string) {
+	e.mu.RLock()
+	site, exists := e.liveState[siteID]
+	e.mu.RUnlock()
+	if !exists {
+		return
+	}
+
+	staleAfter := time.Duration(site.Interval) * time.Second * 3
+	if staleAfter < time.Minute {
+		staleAfter = time.Minute
+	}
+
+	now := time.Now()
+	e.probeResultsMu.Lock()
+	if e.probeResults[siteID] == nil {
+		e.probeResults[siteID] = make(map[string]NodeResult)
+	}
+	e.probeResults[siteID][nodeID] = NodeResult{
+		NodeID:      nodeID,
+		IsUp:        isUp,
+		LatencyNs:   latencyNs,
+		CheckedAt:   now,
+		ErrorReason: errorReason,
+	}
+	results := make([]NodeResult, 0, len(e.probeResults[siteID]))
+	for id, r := range e.probeResults[siteID] {
+		if now.Sub(r.CheckedAt) > staleAfter {
+			delete(e.probeResults[siteID], id)
+			continue
+		}
+		results = append(results, r)
+	}
+	e.probeResultsMu.Unlock()
+
+	aggUp, avgLatency := AggregateStatus(results, e.aggStrategy)
+
+	probeStatus := models.StatusUp
+	if !aggUp {
+		probeStatus = models.StatusDown
+	}
+
+	updatedSite := site
+	updatedSite.Latency = time.Duration(avgLatency)
+	updatedSite.LastCheck = time.Now()
+	e.handleStatusChange(updatedSite, string(probeStatus), 0, time.Duration(avgLatency), errorReason)
+}
+
+func (e *Engine) GetProbeResults(siteID int) map[string]NodeResult {
+	e.probeResultsMu.RLock()
+	defer e.probeResultsMu.RUnlock()
+	src := e.probeResults[siteID]
+	cp := make(map[string]NodeResult, len(src))
+	for k, v := range src {
+		cp[k] = v
+	}
+	return cp
+}
+
+func (e *Engine) GetStateChanges(siteID int, limit int) []models.StateChange {
+	changes, err := e.db.GetStateChanges(context.Background(), siteID, limit)
+	if err != nil {
+		return nil
+	}
+	return changes
+}
+
+func (e *Engine) GetStateChangesSince(siteID int, since time.Time) []models.StateChange {
+	changes, err := e.db.GetStateChangesSince(context.Background(), siteID, since)
+	if err != nil {
+		return nil
+	}
+	return changes
+}

internal/monitor/safedial.go

@@ -7,8 +7,6 @@ import (
 	"time"
 )
 
-const dialTimeout = 10 * time.Second
-
 var privateRanges []*net.IPNet
 
 func init() {
@@ -62,7 +60,7 @@ func SafeDialContext(allowPrivate bool) func(ctx context.Context, network, addr
 			}
 		}
 
-		dialer := &net.Dialer{Timeout: dialTimeout}
+		dialer := &net.Dialer{Timeout: 10 * time.Second}
 		for _, ip := range ips {
 			target := net.JoinHostPort(ip.IP.String(), port)
 			conn, err := dialer.DialContext(ctx, network, target)

internal/monitor/sites.go

@@ -1,158 +0,0 @@
-package monitor
-
-import (
-	"context"
-	"fmt"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-func (e *Engine) SetActive(active bool) {
-	e.activeMu.Lock()
-	defer e.activeMu.Unlock()
-	if e.isActive != active {
-		e.isActive = active
-		status := "RESUMED (Active)"
-		if !active {
-			status = "PAUSED (Passive)"
-		}
-		e.AddLog(fmt.Sprintf("Engine %s", status))
-	}
-}
-
-func (e *Engine) IsActive() bool {
-	e.activeMu.RLock()
-	defer e.activeMu.RUnlock()
-	return e.isActive
-}
-
-func (e *Engine) GetAllSites() []models.Site {
-	e.mu.RLock()
-	defer e.mu.RUnlock()
-	sites := make([]models.Site, 0, len(e.liveState))
-	for _, s := range e.liveState {
-		sites = append(sites, s)
-	}
-	return sites
-}
-
-func (e *Engine) GetLiveState() map[int]models.Site {
-	e.mu.RLock()
-	defer e.mu.RUnlock()
-	cp := make(map[int]models.Site, len(e.liveState))
-	for k, v := range e.liveState {
-		cp[k] = v
-	}
-	return cp
-}
-
-func (e *Engine) addToTokenIndex(site models.Site) {
-	if site.Type == "push" && site.Token != "" {
-		e.tokenIndex[site.Token] = site.ID
-	}
-}
-
-func (e *Engine) removeFromTokenIndex(id int) {
-	for token, sid := range e.tokenIndex {
-		if sid == id {
-			delete(e.tokenIndex, token)
-			return
-		}
-	}
-}
-
-func (e *Engine) UpdateSiteConfig(cfg models.SiteConfig) {
-	e.mu.Lock()
-	if existing, ok := e.liveState[cfg.ID]; ok {
-		e.removeFromTokenIndex(cfg.ID)
-		existing.SiteConfig = cfg
-		e.liveState[cfg.ID] = existing
-		e.addToTokenIndex(existing)
-	}
-	e.mu.Unlock()
-
-	e.signalRecheck(cfg.ID)
-}
-
-func (e *Engine) RemoveSite(id int) {
-	e.mu.Lock()
-	e.removeFromTokenIndex(id)
-	delete(e.liveState, id)
-	e.mu.Unlock()
-	e.removeHistory(id)
-
-	e.probeResultsMu.Lock()
-	delete(e.probeResults, id)
-	e.probeResultsMu.Unlock()
-
-	e.recheckMu.Lock()
-	delete(e.recheck, id)
-	e.recheckMu.Unlock()
-}
-
-func (e *Engine) ToggleSitePause(id int) bool {
-	var (
-		paused bool
-		name   string
-	)
-	_, ok := e.applyState(id, func(s *models.Site) {
-		s.Paused = !s.Paused
-		paused = s.Paused
-		name = s.Name
-	})
-	if !ok {
-		return false
-	}
-	if paused {
-		e.AddLog(fmt.Sprintf("Monitor '%s' paused", name))
-	} else {
-		e.AddLog(fmt.Sprintf("Monitor '%s' resumed", name))
-	}
-	return paused
-}
-
-// applyState atomically reads, mutates, and writes back the live entry for id.
-// The mutator runs under the engine write lock and receives a pointer to the
-// CURRENT live state, so concurrent config edits, pauses, and heartbeats are
-// never clobbered by a stale snapshot. The mutator must only touch runtime /
-// check-result fields — config fields (Name/URL/Type/Token/Interval/AlertID/…)
-// are owned by UpdateSiteConfig and must not be written here. Returns the
-// post-mutation copy and whether the site still exists.
-func (e *Engine) applyState(id int, mutate func(s *models.Site)) (models.Site, bool) {
-	e.mu.Lock()
-	defer e.mu.Unlock()
-	cur, ok := e.liveState[id]
-	if !ok {
-		return models.Site{}, false
-	}
-	mutate(&cur)
-	e.liveState[id] = cur
-	return cur, true
-}
-
-func (e *Engine) GetDisplayStatus(site models.Site) string {
-	if site.Paused {
-		return "PAUSED"
-	}
-	if e.isInMaintenance(site.ID) {
-		return "MAINT"
-	}
-	return string(site.Status)
-}
-
-func (e *Engine) GetStateChanges(ctx context.Context, siteID int, limit int) []models.StateChange {
-	changes, err := e.db.GetStateChanges(ctx, siteID, limit)
-	if err != nil {
-		return nil
-	}
-	return changes
-}
-
-func (e *Engine) GetStateChangesSince(ctx context.Context, siteID int, since time.Time) []models.StateChange {
-	changes, err := e.db.GetStateChangesSince(ctx, siteID, since)
-	if err != nil {
-		return nil
-	}
-	return changes
-}

internal/monitor/sites_test.go

@@ -1,133 +0,0 @@
-package monitor
-
-import (
-	"testing"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-func TestUpdateSiteConfig_PreservesRuntime(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", URL: "http://old.com"},
-		SiteState:  models.SiteState{Status: "DOWN", FailureCount: 3, Latency: 100 * time.Millisecond},
-	}
-	injectSite(e, site)
-
-	updated := models.SiteConfig{ID: 1, Name: "test", URL: "http://new.com", Interval: 60}
-	e.UpdateSiteConfig(updated)
-
-	s, _ := getSite(e, 1)
-	if s.URL != "http://new.com" {
-		t.Errorf("expected URL updated, got %s", s.URL)
-	}
-	if s.Status != "DOWN" {
-		t.Errorf("expected Status preserved, got %s", s.Status)
-	}
-	if s.FailureCount != 3 {
-		t.Errorf("expected FailureCount preserved, got %d", s.FailureCount)
-	}
-	if s.Latency != 100*time.Millisecond {
-		t.Errorf("expected Latency preserved, got %v", s.Latency)
-	}
-}
-
-func TestRemoveSite_CleansUp(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test", Type: "push", Token: "tok1"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-	e.recordCheck(1, 5*time.Millisecond, true)
-
-	e.RemoveSite(1)
-
-	if _, ok := getSite(e, 1); ok {
-		t.Error("expected site removed from liveState")
-	}
-	if e.RecordHeartbeat("tok1") {
-		t.Error("expected token removed from index")
-	}
-	if _, ok := e.GetHistory(1); ok {
-		t.Error("expected history removed")
-	}
-}
-
-func TestToggleSitePause(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	site := models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "test"},
-		SiteState:  models.SiteState{Status: "UP"},
-	}
-	injectSite(e, site)
-
-	paused := e.ToggleSitePause(1)
-	if !paused {
-		t.Error("expected paused=true after first toggle")
-	}
-	s, _ := getSite(e, 1)
-	if !s.Paused {
-		t.Error("expected Paused=true in state")
-	}
-
-	paused = e.ToggleSitePause(1)
-	if paused {
-		t.Error("expected paused=false after second toggle")
-	}
-}
-
-func TestToggleSitePause_NonexistentSite(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	if e.ToggleSitePause(999) {
-		t.Error("expected false for nonexistent site")
-	}
-}
-
-func TestGetAllSites_ReturnsCopy(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	injectSite(e, models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "s1"},
-		SiteState:  models.SiteState{Status: "UP"},
-	})
-	injectSite(e, models.Site{
-		SiteConfig: models.SiteConfig{ID: 2, Name: "s2"},
-		SiteState:  models.SiteState{Status: "DOWN"},
-	})
-
-	sites := e.GetAllSites()
-	if len(sites) != 2 {
-		t.Fatalf("expected 2 sites, got %d", len(sites))
-	}
-	sites[0].Name = "mutated"
-
-	fresh := e.GetAllSites()
-	for _, s := range fresh {
-		if s.Name == "mutated" {
-			t.Error("GetAllSites returned reference, not copy")
-		}
-	}
-}
-
-func TestGetLiveState_ReturnsCopy(t *testing.T) {
-	ms := newMockStore()
-	e := newTestEngine(ms)
-	injectSite(e, models.Site{
-		SiteConfig: models.SiteConfig{ID: 1, Name: "s1"},
-		SiteState:  models.SiteState{Status: "UP"},
-	})
-
-	state := e.GetLiveState()
-	state[1] = models.Site{SiteConfig: models.SiteConfig{Name: "mutated"}}
-
-	fresh := e.GetLiveState()
-	if fresh[1].Name == "mutated" {
-		t.Error("GetLiveState returned reference, not copy")
-	}
-}

internal/server/ratelimit.go

@@ -14,11 +14,6 @@ import (
 // guard.
 const maxVisitors = 10000
 
-const (
-	visitorCleanupInterval = 5 * time.Minute
-	visitorIdleCutoff      = 10 * time.Minute
-)
-
 type visitor struct {
 	tokens   float64
 	lastSeen time.Time
@@ -95,13 +90,13 @@ func (rl *RateLimiter) evictOldest() {
 }
 
 func (rl *RateLimiter) cleanup() {
-	ticker := time.NewTicker(visitorCleanupInterval)
+	ticker := time.NewTicker(5 * time.Minute)
 	defer ticker.Stop()
 	for {
 		select {
 		case <-ticker.C:
 			rl.mu.Lock()
-			cutoff := time.Now().Add(-visitorIdleCutoff)
+			cutoff := time.Now().Add(-10 * time.Minute)
 			for ip, v := range rl.visitors {
 				if v.lastSeen.Before(cutoff) {
 					delete(rl.visitors, ip)

internal/server/server.go

@@ -45,27 +45,15 @@ type Server struct {
 	statusRL *RateLimiter
 }
 
-const (
-	pushRateLimit   = 60
-	probeRateLimit  = 30
-	backupRateLimit = 10
-	statusRateLimit = 120
-
-	httpReadHeaderTimeout = 10 * time.Second
-	httpReadTimeout       = 30 * time.Second
-	httpWriteTimeout      = 60 * time.Second
-	httpIdleTimeout       = 120 * time.Second
-)
-
 func NewServer(cfg ServerConfig, s store.Store, eng *monitor.Engine) *Server {
 	return &Server{
 		cfg:      cfg,
 		store:    s,
 		eng:      eng,
-		pushRL:   NewRateLimiter(pushRateLimit, cfg.TrustedProxies),
-		probeRL:  NewRateLimiter(probeRateLimit, cfg.TrustedProxies),
-		backupRL: NewRateLimiter(backupRateLimit, cfg.TrustedProxies),
-		statusRL: NewRateLimiter(statusRateLimit, cfg.TrustedProxies),
+		pushRL:   NewRateLimiter(60, cfg.TrustedProxies),
+		probeRL:  NewRateLimiter(30, cfg.TrustedProxies),
+		backupRL: NewRateLimiter(10, cfg.TrustedProxies),
+		statusRL: NewRateLimiter(120, cfg.TrustedProxies),
 	}
 }
 
@@ -89,10 +77,10 @@ func (s *Server) Start() *http.Server {
 	httpSrv := &http.Server{
 		Addr:              addr,
 		Handler:           handler,
-		ReadHeaderTimeout: httpReadHeaderTimeout,
-		ReadTimeout:       httpReadTimeout,
-		WriteTimeout:      httpWriteTimeout,
-		IdleTimeout:       httpIdleTimeout,
+		ReadHeaderTimeout: 10 * time.Second,
+		ReadTimeout:       30 * time.Second,
+		WriteTimeout:      60 * time.Second,
+		IdleTimeout:       120 * time.Second,
 	}
 	go func() {
 		if s.cfg.TLSCert != "" && s.cfg.TLSKey != "" {
@@ -139,7 +127,7 @@ func (s *Server) routes() http.Handler {
 }
 
 func (s *Server) requireAuth(r *http.Request) bool {
-	return s.cfg.ClusterKey != "" && checkSecret(r.Header.Get("X-Uptop-Secret"), s.cfg.ClusterKey)
+	return s.cfg.ClusterKey != "" && checkSecret(r.Header.Get("X-Upkeep-Secret"), s.cfg.ClusterKey)
 }
 
 func (s *Server) handlePush(w http.ResponseWriter, r *http.Request) {
@@ -171,6 +159,10 @@ func (s *Server) handleHealth(w http.ResponseWriter, r *http.Request) {
 		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
 		return
 	}
+	if s.cfg.ClusterKey != "" && !checkSecret(r.Header.Get("X-Upkeep-Secret"), s.cfg.ClusterKey) {
+		http.Error(w, "Unauthorized", http.StatusUnauthorized)
+		return
+	}
 	w.WriteHeader(http.StatusOK)
 	_, _ = w.Write([]byte("OK"))
 }

internal/server/server_test.go

@@ -77,12 +77,8 @@ func (m *mockStore) GetActiveMaintenanceWindows(_ context.Context) ([]models.Mai
 
 // --- Helpers ---
 
-func freePort(t *testing.T) int {
-	t.Helper()
-	ln, err := net.Listen("tcp", "127.0.0.1:0")
-	if err != nil {
-		t.Fatalf("freePort: %v", err)
-	}
+func freePort() int {
+	ln, _ := net.Listen("tcp", "127.0.0.1:0")
 	port := ln.Addr().(*net.TCPAddr).Port
 	ln.Close()
 	return port
@@ -99,7 +95,7 @@ func newTestServer(t *testing.T, clusterKey string, enableStatus bool) *testServ
 	t.Helper()
 	ms := newMockStore()
 	eng := monitor.NewEngine(ms)
-	port := freePort(t)
+	port := freePort()
 
 	srv := Start(ServerConfig{
 		Port:         port,
@@ -145,7 +141,7 @@ func authReq(method, url, secret string, body []byte) (*http.Response, error) {
 		return nil, err
 	}
 	if secret != "" {
-		req.Header.Set("X-Uptop-Secret", secret)
+		req.Header.Set("X-Upkeep-Secret", secret)
 	}
 	return http.DefaultClient.Do(req)
 }
@@ -223,8 +219,8 @@ func TestHealth_WrongSecret(t *testing.T) {
 		t.Fatal(err)
 	}
 	defer resp.Body.Close()
-	if resp.StatusCode != 200 {
-		t.Errorf("health is unauthenticated, expected 200, got %d", resp.StatusCode)
+	if resp.StatusCode != 401 {
+		t.Errorf("expected 401, got %d", resp.StatusCode)
 	}
 }
 
@@ -289,10 +285,7 @@ func TestImport_MethodNotAllowed(t *testing.T) {
 
 func TestImport_Unauthorized(t *testing.T) {
 	ts := newTestServer(t, "secret", false)
-	body, err := json.Marshal(models.Backup{})
-	if err != nil {
-		t.Fatalf("json.Marshal: %v", err)
-	}
+	body, _ := json.Marshal(models.Backup{})
 	resp, err := authReq("POST", ts.baseURL+"/api/backup/import", "wrong", body)
 	if err != nil {
 		t.Fatal(err)
@@ -308,10 +301,7 @@ func TestImport_Success(t *testing.T) {
 	backup := models.Backup{
 		Sites: []models.SiteConfig{{Name: "imported", URL: "http://example.com"}},
 	}
-	body, err := json.Marshal(backup)
-	if err != nil {
-		t.Fatalf("json.Marshal: %v", err)
-	}
+	body, _ := json.Marshal(backup)
 	resp, err := authReq("POST", ts.baseURL+"/api/backup/import", "secret", body)
 	if err != nil {
 		t.Fatal(err)
@@ -343,12 +333,9 @@ func TestImport_InvalidJSON(t *testing.T) {
 
 func TestProbeRegister_Success(t *testing.T) {
 	ts := newTestServer(t, "secret", false)
-	body, err := json.Marshal(map[string]string{
+	body, _ := json.Marshal(map[string]string{
 		"id": "node-1", "name": "US East", "region": "us-east",
 	})
-	if err != nil {
-		t.Fatalf("json.Marshal: %v", err)
-	}
 	resp, err := authReq("POST", ts.baseURL+"/api/probe/register", "secret", body)
 	if err != nil {
 		t.Fatal(err)
@@ -369,10 +356,7 @@ func TestProbeRegister_Success(t *testing.T) {
 
 func TestProbeRegister_MissingID(t *testing.T) {
 	ts := newTestServer(t, "secret", false)
-	body, err := json.Marshal(map[string]string{"name": "test"})
-	if err != nil {
-		t.Fatalf("json.Marshal: %v", err)
-	}
+	body, _ := json.Marshal(map[string]string{"name": "test"})
 	resp, err := authReq("POST", ts.baseURL+"/api/probe/register", "secret", body)
 	if err != nil {
 		t.Fatal(err)
@@ -385,10 +369,7 @@ func TestProbeRegister_MissingID(t *testing.T) {
 
 func TestProbeRegister_Unauthorized(t *testing.T) {
 	ts := newTestServer(t, "secret", false)
-	body, err := json.Marshal(map[string]string{"id": "node-1"})
-	if err != nil {
-		t.Fatalf("json.Marshal: %v", err)
-	}
+	body, _ := json.Marshal(map[string]string{"id": "node-1"})
 	resp, err := authReq("POST", ts.baseURL+"/api/probe/register", "wrong", body)
 	if err != nil {
 		t.Fatal(err)
@@ -403,15 +384,12 @@ func TestProbeRegister_Unauthorized(t *testing.T) {
 
 func TestProbeResults_Success(t *testing.T) {
 	ts := newTestServer(t, "secret", false)
-	body, err := json.Marshal(map[string]any{
+	body, _ := json.Marshal(map[string]any{
 		"node_id": "node-1",
 		"results": []map[string]any{
 			{"site_id": 1, "latency_ns": 5000000, "is_up": true},
 		},
 	})
-	if err != nil {
-		t.Fatalf("json.Marshal: %v", err)
-	}
 	resp, err := authReq("POST", ts.baseURL+"/api/probe/results", "secret", body)
 	if err != nil {
 		t.Fatal(err)
@@ -424,12 +402,9 @@ func TestProbeResults_Success(t *testing.T) {
 
 func TestProbeResults_MissingNodeID(t *testing.T) {
 	ts := newTestServer(t, "secret", false)
-	body, err := json.Marshal(map[string]any{
+	body, _ := json.Marshal(map[string]any{
 		"results": []map[string]any{},
 	})
-	if err != nil {
-		t.Fatalf("json.Marshal: %v", err)
-	}
 	resp, err := authReq("POST", ts.baseURL+"/api/probe/results", "secret", body)
 	if err != nil {
 		t.Fatal(err)
@@ -598,10 +573,7 @@ func TestClientIP_TrustedProxyHandling(t *testing.T) {
 	}
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			r, err := http.NewRequest(http.MethodGet, "/", nil)
-			if err != nil {
-				t.Fatalf("http.NewRequest: %v", err)
-			}
+			r, _ := http.NewRequest(http.MethodGet, "/", nil)
 			r.RemoteAddr = tt.remoteAddr
 			if tt.xff != "" {
 				r.Header.Set("X-Forwarded-For", tt.xff)
@@ -619,10 +591,7 @@ func TestRateLimit_SpoofedXFFCannotBypass(t *testing.T) {
 	rl := NewRateLimiter(60, nil) // no trusted proxies
 	allowed := 0
 	for i := 0; i < 200; i++ {
-		r, err := http.NewRequest(http.MethodGet, "/", nil)
-		if err != nil {
-			t.Fatalf("http.NewRequest: %v", err)
-		}
+		r, _ := http.NewRequest(http.MethodGet, "/", nil)
 		r.RemoteAddr = "203.0.113.9:5000"
 		r.Header.Set("X-Forwarded-For", fmt.Sprintf("9.9.9.%d", i%256))
 		if rl.Allow(clientIP(r, rl.trusted)) {

internal/store/crypto_test.go

@@ -75,14 +75,8 @@ func TestEncryptorUniqueCiphertexts(t *testing.T) {
 		t.Fatal(err)
 	}
 
-	a, err := enc.Encrypt("same")
-	if err != nil {
-		t.Fatalf("first encrypt: %v", err)
-	}
-	b, err := enc.Encrypt("same")
-	if err != nil {
-		t.Fatalf("second encrypt: %v", err)
-	}
+	a, _ := enc.Encrypt("same")
+	b, _ := enc.Encrypt("same")
 	if a == b {
 		t.Error("two encryptions of same plaintext should produce different ciphertexts")
 	}

internal/store/sqlstore.go

@@ -5,8 +5,8 @@ import (
 	"crypto/rand"
 	"database/sql"
 	"encoding/hex"
+	"encoding/json"
 	"fmt"
-	"strings"
 	"time"
 
 	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
@@ -17,10 +17,6 @@ const (
 	maxLogRows             = 200
 	maxStateChangesPerSite = 5000
 	maxMaintenanceExport   = 1000
-	maxOpenConns           = 25
-	maxIdleConns           = 5
-	connMaxLifetime        = 5 * time.Minute
-	tokenByteLen           = 16
 )
 
 type SQLStore struct {
@@ -35,9 +31,9 @@ func NewSQLStore(driverName, dsn string, dialect Dialect) (*SQLStore, error) {
 	if err != nil {
 		return nil, err
 	}
-	db.SetMaxOpenConns(maxOpenConns)
-	db.SetMaxIdleConns(maxIdleConns)
-	db.SetConnMaxLifetime(connMaxLifetime)
+	db.SetMaxOpenConns(25)
+	db.SetMaxIdleConns(5)
+	db.SetConnMaxLifetime(5 * time.Minute)
 	_, isDollar := dialect.(*PostgresDialect)
 	return &SQLStore{db: db, dialect: dialect, dollar: isDollar}, nil
 }
@@ -65,7 +61,7 @@ func (s *SQLStore) q(query string) string {
 }
 
 func generateToken() (string, error) {
-	b := make([]byte, tokenByteLen)
+	b := make([]byte, 16)
 	if _, err := rand.Read(b); err != nil {
 		return "", fmt.Errorf("crypto/rand failed: %w", err)
 	}
@@ -217,6 +213,40 @@ func (s *SQLStore) GetSiteByName(ctx context.Context, name string) (models.SiteC
 	return st, err
 }
 
+func (s *SQLStore) unmarshalSettings(raw string) (map[string]string, error) {
+	decrypted, err := s.decryptSettings(raw)
+	if err != nil {
+		return nil, fmt.Errorf("decrypt settings: %w", err)
+	}
+	var m map[string]string
+	if err := json.Unmarshal([]byte(decrypted), &m); err != nil {
+		return nil, fmt.Errorf("unmarshal settings: %w", err)
+	}
+	return m, nil
+}
+
+func (s *SQLStore) marshalSettings(settings map[string]string) (string, error) {
+	jsonBytes, err := json.Marshal(settings)
+	if err != nil {
+		return "", err
+	}
+	return s.encryptSettings(string(jsonBytes))
+}
+
+func (s *SQLStore) GetAlertByName(ctx context.Context, name string) (models.AlertConfig, error) {
+	var a models.AlertConfig
+	var settingsRaw string
+	err := s.db.QueryRowContext(ctx, s.q("SELECT id, name, type, settings FROM alerts WHERE name = ?"), name).Scan(&a.ID, &a.Name, &a.Type, &settingsRaw)
+	if err != nil {
+		return a, err
+	}
+	a.Settings, err = s.unmarshalSettings(settingsRaw)
+	if err != nil {
+		return a, fmt.Errorf("alert %q: %w", name, err)
+	}
+	return a, nil
+}
+
 func (s *SQLStore) AddSiteReturningID(ctx context.Context, site models.SiteConfig) (int, error) {
 	token := ""
 	if site.Type == "push" {
@@ -243,6 +273,89 @@ func (s *SQLStore) AddSiteReturningID(ctx context.Context, site models.SiteConfi
 	return int(id), err
 }
 
+func (s *SQLStore) AddAlertReturningID(ctx context.Context, name, aType string, settings map[string]string) (int, error) {
+	stored, err := s.marshalSettings(settings)
+	if err != nil {
+		return 0, err
+	}
+	if s.dollar {
+		var id int
+		err := s.db.QueryRowContext(ctx, s.q("INSERT INTO alerts (name, type, settings) VALUES (?, ?, ?) RETURNING id"), name, aType, stored).Scan(&id)
+		return id, err
+	}
+	result, err := s.db.ExecContext(ctx, s.q("INSERT INTO alerts (name, type, settings) VALUES (?, ?, ?)"), name, aType, stored)
+	if err != nil {
+		return 0, err
+	}
+	id, err := result.LastInsertId()
+	return int(id), err
+}
+
+func (s *SQLStore) GetAllAlerts(ctx context.Context) ([]models.AlertConfig, error) {
+	rows, err := s.db.QueryContext(ctx, "SELECT id, name, type, settings FROM alerts")
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+	var alerts []models.AlertConfig
+	for rows.Next() {
+		var a models.AlertConfig
+		var settingsRaw string
+		if err := rows.Scan(&a.ID, &a.Name, &a.Type, &settingsRaw); err != nil {
+			return alerts, err
+		}
+		a.Settings, err = s.unmarshalSettings(settingsRaw)
+		if err != nil {
+			return alerts, fmt.Errorf("alert %q: %w", a.Name, err)
+		}
+		alerts = append(alerts, a)
+	}
+	return alerts, rows.Err()
+}
+
+func (s *SQLStore) GetAlert(ctx context.Context, id int) (models.AlertConfig, error) {
+	var a models.AlertConfig
+	var settingsRaw string
+	err := s.db.QueryRowContext(ctx, s.q("SELECT id, name, type, settings FROM alerts WHERE id = ?"), id).Scan(&a.ID, &a.Name, &a.Type, &settingsRaw)
+	if err != nil {
+		return a, err
+	}
+	a.Settings, err = s.unmarshalSettings(settingsRaw)
+	if err != nil {
+		return a, fmt.Errorf("alert %d: %w", id, err)
+	}
+	return a, nil
+}
+
+func (s *SQLStore) AddAlert(ctx context.Context, name, aType string, settings map[string]string) error {
+	stored, err := s.marshalSettings(settings)
+	if err != nil {
+		return err
+	}
+	_, err = s.db.ExecContext(ctx, s.q("INSERT INTO alerts (name, type, settings) VALUES (?, ?, ?)"), name, aType, stored)
+	return err
+}
+
+func (s *SQLStore) UpdateAlert(ctx context.Context, id int, name, aType string, settings map[string]string) error {
+	stored, err := s.marshalSettings(settings)
+	if err != nil {
+		return err
+	}
+	_, err = s.db.ExecContext(ctx, s.q("UPDATE alerts SET name=?, type=?, settings=? WHERE id=?"), name, aType, stored, id)
+	return err
+}
+
+func (s *SQLStore) DeleteAlert(ctx context.Context, id int) error {
+	if _, err := s.db.ExecContext(ctx, s.q("UPDATE sites SET alert_id = 0 WHERE alert_id = ?"), id); err != nil {
+		return err
+	}
+	if _, err := s.db.ExecContext(ctx, s.q("DELETE FROM alerts WHERE id=?"), id); err != nil {
+		return err
+	}
+	s.dialect.ResetSequenceOnEmpty(s.db, "alerts")
+	return nil
+}
+
 func (s *SQLStore) GetAllUsers(ctx context.Context) ([]models.User, error) {
 	rows, err := s.db.QueryContext(ctx, "SELECT id, username, public_key, role FROM users")
 	if err != nil {
@@ -275,6 +388,85 @@ func (s *SQLStore) DeleteUser(ctx context.Context, id int) error {
 	return err
 }
 
+func (s *SQLStore) SaveStateChange(ctx context.Context, siteID int, fromStatus, toStatus, errorReason string) error {
+	_, err := s.db.ExecContext(ctx, s.q("INSERT INTO state_changes (site_id, from_status, to_status, error_reason) VALUES (?, ?, ?, ?)"),
+		siteID, fromStatus, toStatus, errorReason)
+	return err
+}
+
+func (s *SQLStore) GetStateChanges(ctx context.Context, siteID int, limit int) ([]models.StateChange, error) {
+	rows, err := s.db.QueryContext(ctx, s.q("SELECT id, site_id, from_status, to_status, error_reason, changed_at FROM state_changes WHERE site_id = ? ORDER BY changed_at DESC LIMIT ?"), siteID, limit)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+	var changes []models.StateChange
+	for rows.Next() {
+		var sc models.StateChange
+		if err := rows.Scan(&sc.ID, &sc.SiteID, &sc.FromStatus, &sc.ToStatus, &sc.ErrorReason, &sc.ChangedAt); err != nil {
+			return changes, err
+		}
+		changes = append(changes, sc)
+	}
+	return changes, rows.Err()
+}
+
+func (s *SQLStore) GetStateChangesSince(ctx context.Context, siteID int, since time.Time) ([]models.StateChange, error) {
+	rows, err := s.db.QueryContext(ctx, s.q("SELECT id, site_id, from_status, to_status, error_reason, changed_at FROM state_changes WHERE site_id = ? AND changed_at >= ? ORDER BY changed_at DESC"), siteID, since)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+	var changes []models.StateChange
+	for rows.Next() {
+		var sc models.StateChange
+		if err := rows.Scan(&sc.ID, &sc.SiteID, &sc.FromStatus, &sc.ToStatus, &sc.ErrorReason, &sc.ChangedAt); err != nil {
+			return changes, err
+		}
+		changes = append(changes, sc)
+	}
+	return changes, rows.Err()
+}
+
+func (s *SQLStore) SaveCheck(ctx context.Context, siteID int, latencyNs int64, isUp bool) error {
+	return s.SaveCheckFromNode(ctx, siteID, "", latencyNs, isUp)
+}
+
+// SaveCheckFromNode inserts a single check row. Retention is handled out of
+// band by PruneCheckHistory on a timer, not per-insert, to keep the write hot
+// path a plain INSERT.
+func (s *SQLStore) SaveCheckFromNode(ctx context.Context, siteID int, nodeID string, latencyNs int64, isUp bool) error {
+	_, err := s.db.ExecContext(ctx, s.q("INSERT INTO check_history (site_id, node_id, latency_ns, is_up) VALUES (?, ?, ?, ?)"), siteID, nodeID, latencyNs, isUp)
+	return err
+}
+
+// PruneCheckHistory trims check_history to the newest maxCheckHistory rows per
+// site, across all sites, in one pass. Intended to run periodically.
+func (s *SQLStore) PruneCheckHistory(ctx context.Context) error {
+	q := fmt.Sprintf(`DELETE FROM check_history WHERE id IN (
+		SELECT id FROM (
+			SELECT id, ROW_NUMBER() OVER (PARTITION BY site_id ORDER BY checked_at DESC, id DESC) AS rn
+			FROM check_history
+		) ranked WHERE rn > %d
+	)`, maxCheckHistory)
+	_, err := s.db.ExecContext(ctx, s.q(q))
+	return err
+}
+
+// PruneStateChanges trims state_changes to the newest maxStateChangesPerSite
+// rows per site. Generous so realistic SLA windows are unaffected; bounds the
+// otherwise unbounded growth of a flapping monitor's history.
+func (s *SQLStore) PruneStateChanges(ctx context.Context) error {
+	q := fmt.Sprintf(`DELETE FROM state_changes WHERE id IN (
+		SELECT id FROM (
+			SELECT id, ROW_NUMBER() OVER (PARTITION BY site_id ORDER BY changed_at DESC, id DESC) AS rn
+			FROM state_changes
+		) ranked WHERE rn > %d
+	)`, maxStateChangesPerSite)
+	_, err := s.db.ExecContext(ctx, s.q(q))
+	return err
+}
+
 func (s *SQLStore) RegisterNode(ctx context.Context, node models.ProbeNode) error {
 	_, err := s.db.ExecContext(ctx, s.dialect.UpsertNodeSQL(), node.ID, node.Name, node.Region, node.Version)
 	return err
@@ -362,27 +554,145 @@ func (s *SQLStore) PruneLogs(ctx context.Context) error {
 	return err
 }
 
-func (s *SQLStore) LoadLogs(ctx context.Context, limit int) ([]models.LogEntry, error) {
-	rows, err := s.db.QueryContext(ctx, s.q("SELECT message, created_at FROM logs ORDER BY created_at DESC, id DESC LIMIT ?"), limit)
+func (s *SQLStore) LoadLogs(ctx context.Context, limit int) ([]string, error) {
+	rows, err := s.db.QueryContext(ctx, s.q("SELECT message FROM logs ORDER BY created_at DESC LIMIT ?"), limit)
 	if err != nil {
 		return nil, err
 	}
 	defer rows.Close()
-	var entries []models.LogEntry
+	var logs []string
 	for rows.Next() {
-		var e models.LogEntry
-		if err := rows.Scan(&e.Message, &e.CreatedAt); err != nil {
-			return entries, err
+		var msg string
+		if err := rows.Scan(&msg); err != nil {
+			return logs, err
 		}
-		// Strip legacy [HH:MM] or [HH:MM:SS] prefix from pre-refactor entries.
-		if len(e.Message) > 3 && e.Message[0] == '[' {
-			if idx := strings.Index(e.Message, "]"); idx > 0 && idx < 12 {
-				e.Message = strings.TrimSpace(e.Message[idx+1:])
-			}
-		}
-		entries = append(entries, e)
+		logs = append(logs, msg)
 	}
-	return entries, rows.Err()
+	return logs, rows.Err()
+}
+
+func (s *SQLStore) LoadAllHistory(ctx context.Context, limit int) (map[int][]models.CheckRecord, error) {
+	result := make(map[int][]models.CheckRecord)
+	rows, err := s.db.QueryContext(ctx, s.q(`
+		SELECT site_id, latency_ns, is_up FROM (
+			SELECT site_id, latency_ns, is_up,
+				ROW_NUMBER() OVER (PARTITION BY site_id ORDER BY checked_at DESC) AS rn
+			FROM check_history
+		) sub WHERE rn <= ?`), limit)
+	if err != nil {
+		return result, err
+	}
+	defer rows.Close()
+	for rows.Next() {
+		var r models.CheckRecord
+		if err := rows.Scan(&r.SiteID, &r.LatencyNs, &r.IsUp); err != nil {
+			return result, err
+		}
+		result[r.SiteID] = append(result[r.SiteID], r)
+	}
+	for id, records := range result {
+		for i, j := 0, len(records)-1; i < j; i, j = i+1, j-1 {
+			records[i], records[j] = records[j], records[i]
+		}
+		result[id] = records
+	}
+	return result, rows.Err()
+}
+
+func (s *SQLStore) scanMaintenanceWindow(rows *sql.Rows) (models.MaintenanceWindow, error) {
+	var mw models.MaintenanceWindow
+	var endTime sql.NullTime
+	if err := rows.Scan(&mw.ID, &mw.MonitorID, &mw.Title, &mw.Description, &mw.Type, &mw.StartTime, &endTime, &mw.CreatedBy, &mw.CreatedAt); err != nil {
+		return mw, err
+	}
+	if endTime.Valid {
+		mw.EndTime = endTime.Time
+	}
+	return mw, nil
+}
+
+func (s *SQLStore) GetActiveMaintenanceWindows(ctx context.Context) ([]models.MaintenanceWindow, error) {
+	rows, err := s.db.QueryContext(ctx, s.q("SELECT id, monitor_id, title, description, type, start_time, end_time, created_by, created_at FROM maintenance_windows WHERE start_time <= CURRENT_TIMESTAMP AND (end_time IS NULL OR end_time > CURRENT_TIMESTAMP) ORDER BY start_time DESC"))
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+	var windows []models.MaintenanceWindow
+	for rows.Next() {
+		mw, err := s.scanMaintenanceWindow(rows)
+		if err != nil {
+			return windows, err
+		}
+		windows = append(windows, mw)
+	}
+	return windows, rows.Err()
+}
+
+func (s *SQLStore) GetAllMaintenanceWindows(ctx context.Context, limit int) ([]models.MaintenanceWindow, error) {
+	rows, err := s.db.QueryContext(ctx, s.q("SELECT id, monitor_id, title, description, type, start_time, end_time, created_by, created_at FROM maintenance_windows ORDER BY created_at DESC LIMIT ?"), limit)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+	var windows []models.MaintenanceWindow
+	for rows.Next() {
+		mw, err := s.scanMaintenanceWindow(rows)
+		if err != nil {
+			return windows, err
+		}
+		windows = append(windows, mw)
+	}
+	return windows, rows.Err()
+}
+
+func (s *SQLStore) AddMaintenanceWindow(ctx context.Context, mw models.MaintenanceWindow) error {
+	if mw.StartTime.IsZero() {
+		mw.StartTime = time.Now()
+	}
+	_, err := s.db.ExecContext(ctx, s.q("INSERT INTO maintenance_windows (monitor_id, title, description, type, start_time, end_time, created_by) VALUES (?, ?, ?, ?, ?, ?, ?)"),
+		mw.MonitorID, mw.Title, mw.Description, mw.Type, mw.StartTime, sql.NullTime{Time: mw.EndTime, Valid: !mw.EndTime.IsZero()}, mw.CreatedBy)
+	return err
+}
+
+func (s *SQLStore) EndMaintenanceWindow(ctx context.Context, id int) error {
+	_, err := s.db.ExecContext(ctx, s.q("UPDATE maintenance_windows SET end_time = CURRENT_TIMESTAMP WHERE id = ?"), id)
+	return err
+}
+
+func (s *SQLStore) DeleteMaintenanceWindow(ctx context.Context, id int) error {
+	_, err := s.db.ExecContext(ctx, s.q("DELETE FROM maintenance_windows WHERE id = ?"), id)
+	if err != nil {
+		return err
+	}
+	s.dialect.ResetSequenceOnEmpty(s.db, "maintenance_windows")
+	return nil
+}
+
+func (s *SQLStore) PruneExpiredMaintenanceWindows(ctx context.Context, retention time.Duration) (int64, error) {
+	cutoff := time.Now().Add(-retention)
+	result, err := s.db.ExecContext(ctx,
+		s.q("DELETE FROM maintenance_windows WHERE end_time IS NOT NULL AND end_time < ?"),
+		cutoff,
+	)
+	if err != nil {
+		return 0, err
+	}
+	return result.RowsAffected()
+}
+
+func (s *SQLStore) IsMonitorInMaintenance(ctx context.Context, monitorID int) (bool, error) {
+	var count int
+	err := s.db.QueryRowContext(ctx, s.q(`SELECT COUNT(*) FROM maintenance_windows
+		WHERE type = 'maintenance'
+		AND start_time <= CURRENT_TIMESTAMP
+		AND (end_time IS NULL OR end_time > CURRENT_TIMESTAMP)
+		AND (monitor_id = 0 OR monitor_id = ?
+			OR monitor_id IN (SELECT parent_id FROM sites WHERE id = ? AND parent_id > 0))`),
+		monitorID, monitorID).Scan(&count)
+	if err != nil {
+		return false, err
+	}
+	return count > 0, nil
 }
 
 func (s *SQLStore) GetPreference(ctx context.Context, key string) (string, error) {

internal/store/sqlstore_alerts.go

@@ -1,126 +0,0 @@
-package store
-
-import (
-	"context"
-	"encoding/json"
-	"fmt"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-func (s *SQLStore) unmarshalSettings(raw string) (map[string]string, error) {
-	decrypted, err := s.decryptSettings(raw)
-	if err != nil {
-		return nil, fmt.Errorf("decrypt settings: %w", err)
-	}
-	var m map[string]string
-	if err := json.Unmarshal([]byte(decrypted), &m); err != nil {
-		return nil, fmt.Errorf("unmarshal settings: %w", err)
-	}
-	return m, nil
-}
-
-func (s *SQLStore) marshalSettings(settings map[string]string) (string, error) {
-	jsonBytes, err := json.Marshal(settings)
-	if err != nil {
-		return "", err
-	}
-	return s.encryptSettings(string(jsonBytes))
-}
-
-func (s *SQLStore) GetAlertByName(ctx context.Context, name string) (models.AlertConfig, error) {
-	var a models.AlertConfig
-	var settingsRaw string
-	err := s.db.QueryRowContext(ctx, s.q("SELECT id, name, type, settings FROM alerts WHERE name = ?"), name).Scan(&a.ID, &a.Name, &a.Type, &settingsRaw)
-	if err != nil {
-		return a, err
-	}
-	a.Settings, err = s.unmarshalSettings(settingsRaw)
-	if err != nil {
-		return a, fmt.Errorf("alert %q: %w", name, err)
-	}
-	return a, nil
-}
-
-func (s *SQLStore) AddAlertReturningID(ctx context.Context, name, aType string, settings map[string]string) (int, error) {
-	stored, err := s.marshalSettings(settings)
-	if err != nil {
-		return 0, err
-	}
-	if s.dollar {
-		var id int
-		err := s.db.QueryRowContext(ctx, s.q("INSERT INTO alerts (name, type, settings) VALUES (?, ?, ?) RETURNING id"), name, aType, stored).Scan(&id)
-		return id, err
-	}
-	result, err := s.db.ExecContext(ctx, s.q("INSERT INTO alerts (name, type, settings) VALUES (?, ?, ?)"), name, aType, stored)
-	if err != nil {
-		return 0, err
-	}
-	id, err := result.LastInsertId()
-	return int(id), err
-}
-
-func (s *SQLStore) GetAllAlerts(ctx context.Context) ([]models.AlertConfig, error) {
-	rows, err := s.db.QueryContext(ctx, "SELECT id, name, type, settings FROM alerts")
-	if err != nil {
-		return nil, err
-	}
-	defer rows.Close()
-	var alerts []models.AlertConfig
-	for rows.Next() {
-		var a models.AlertConfig
-		var settingsRaw string
-		if err := rows.Scan(&a.ID, &a.Name, &a.Type, &settingsRaw); err != nil {
-			return alerts, err
-		}
-		a.Settings, err = s.unmarshalSettings(settingsRaw)
-		if err != nil {
-			return alerts, fmt.Errorf("alert %q: %w", a.Name, err)
-		}
-		alerts = append(alerts, a)
-	}
-	return alerts, rows.Err()
-}
-
-func (s *SQLStore) GetAlert(ctx context.Context, id int) (models.AlertConfig, error) {
-	var a models.AlertConfig
-	var settingsRaw string
-	err := s.db.QueryRowContext(ctx, s.q("SELECT id, name, type, settings FROM alerts WHERE id = ?"), id).Scan(&a.ID, &a.Name, &a.Type, &settingsRaw)
-	if err != nil {
-		return a, err
-	}
-	a.Settings, err = s.unmarshalSettings(settingsRaw)
-	if err != nil {
-		return a, fmt.Errorf("alert %d: %w", id, err)
-	}
-	return a, nil
-}
-
-func (s *SQLStore) AddAlert(ctx context.Context, name, aType string, settings map[string]string) error {
-	stored, err := s.marshalSettings(settings)
-	if err != nil {
-		return err
-	}
-	_, err = s.db.ExecContext(ctx, s.q("INSERT INTO alerts (name, type, settings) VALUES (?, ?, ?)"), name, aType, stored)
-	return err
-}
-
-func (s *SQLStore) UpdateAlert(ctx context.Context, id int, name, aType string, settings map[string]string) error {
-	stored, err := s.marshalSettings(settings)
-	if err != nil {
-		return err
-	}
-	_, err = s.db.ExecContext(ctx, s.q("UPDATE alerts SET name=?, type=?, settings=? WHERE id=?"), name, aType, stored, id)
-	return err
-}
-
-func (s *SQLStore) DeleteAlert(ctx context.Context, id int) error {
-	if _, err := s.db.ExecContext(ctx, s.q("UPDATE sites SET alert_id = 0 WHERE alert_id = ?"), id); err != nil {
-		return err
-	}
-	if _, err := s.db.ExecContext(ctx, s.q("DELETE FROM alerts WHERE id=?"), id); err != nil {
-		return err
-	}
-	s.dialect.ResetSequenceOnEmpty(s.db, "alerts")
-	return nil
-}

internal/store/sqlstore_alerts_test.go

@@ -1,102 +0,0 @@
-package store
-
-import (
-	"context"
-	"strings"
-	"testing"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-func TestAlertCRUD(t *testing.T) {
-	s := newTestStore(t)
-
-	if err := s.AddAlert(context.Background(), "Discord", "discord", map[string]string{"url": "https://example.com/hook"}); err != nil {
-		t.Fatalf("AddAlert: %v", err)
-	}
-
-	alerts, err := s.GetAllAlerts(context.Background())
-	if err != nil {
-		t.Fatalf("GetAllAlerts: %v", err)
-	}
-	if len(alerts) != 1 {
-		t.Fatalf("expected 1 alert, got %d", len(alerts))
-	}
-	if alerts[0].Type != "discord" {
-		t.Errorf("expected type 'discord', got '%s'", alerts[0].Type)
-	}
-	if alerts[0].Settings["url"] != "https://example.com/hook" {
-		t.Errorf("settings url mismatch")
-	}
-
-	a, err := s.GetAlert(context.Background(), alerts[0].ID)
-	if err != nil {
-		t.Fatalf("GetAlert: %v", err)
-	}
-	if a.Name != "Discord" {
-		t.Errorf("expected name 'Discord', got '%s'", a.Name)
-	}
-
-	if err := s.UpdateAlert(context.Background(), a.ID, "Slack", "slack", map[string]string{"url": "https://slack.com/hook"}); err != nil {
-		t.Fatalf("UpdateAlert: %v", err)
-	}
-
-	a, err = s.GetAlert(context.Background(), a.ID)
-	if err != nil {
-		t.Fatalf("GetAlert: %v", err)
-	}
-	if a.Type != "slack" {
-		t.Errorf("expected type 'slack', got '%s'", a.Type)
-	}
-
-	if err := s.DeleteAlert(context.Background(), a.ID); err != nil {
-		t.Fatalf("DeleteAlert: %v", err)
-	}
-
-	alerts, err = s.GetAllAlerts(context.Background())
-	if err != nil {
-		t.Fatalf("GetAllAlerts: %v", err)
-	}
-	if len(alerts) != 0 {
-		t.Fatalf("expected 0 alerts after delete, got %d", len(alerts))
-	}
-}
-
-// ImportData must encrypt alert settings (like AddAlert/UpdateAlert) so a
-// restore with UPTOP_ENCRYPTION_KEY set never lands secrets in plaintext.
-func TestImportData_EncryptsAlertSettings(t *testing.T) {
-	s := newTestStore(t)
-	enc, err := NewEncryptor(strings.Repeat("ab", 32)) // 64 hex chars = 32 bytes
-	if err != nil {
-		t.Fatalf("NewEncryptor: %v", err)
-	}
-	s.SetEncryptor(enc)
-
-	backup := models.Backup{
-		Alerts: []models.AlertConfig{
-			{ID: 1, Name: "tg", Type: "telegram", Settings: map[string]string{"token": "123:SECRET", "chat_id": "42"}},
-		},
-	}
-	if err := s.ImportData(context.Background(), backup); err != nil {
-		t.Fatalf("ImportData: %v", err)
-	}
-
-	var raw string
-	if err := s.db.QueryRow("SELECT settings FROM alerts WHERE id = 1").Scan(&raw); err != nil {
-		t.Fatalf("query settings: %v", err)
-	}
-	if !strings.HasPrefix(raw, encryptedPrefix) {
-		t.Errorf("imported settings not encrypted: %q", raw)
-	}
-	if strings.Contains(raw, "SECRET") {
-		t.Errorf("plaintext secret found in stored column: %q", raw)
-	}
-
-	alerts, err := s.GetAllAlerts(context.Background())
-	if err != nil {
-		t.Fatalf("GetAllAlerts: %v", err)
-	}
-	if len(alerts) != 1 || alerts[0].Settings["token"] != "123:SECRET" {
-		t.Errorf("decrypt round-trip failed: %+v", alerts)
-	}
-}

internal/store/sqlstore_history.go

@@ -1,116 +0,0 @@
-package store
-
-import (
-	"context"
-	"fmt"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-func (s *SQLStore) SaveStateChange(ctx context.Context, siteID int, fromStatus, toStatus, errorReason string) error {
-	_, err := s.db.ExecContext(ctx, s.q("INSERT INTO state_changes (site_id, from_status, to_status, error_reason) VALUES (?, ?, ?, ?)"),
-		siteID, fromStatus, toStatus, errorReason)
-	return err
-}
-
-func (s *SQLStore) GetStateChanges(ctx context.Context, siteID int, limit int) ([]models.StateChange, error) {
-	rows, err := s.db.QueryContext(ctx, s.q("SELECT id, site_id, from_status, to_status, error_reason, changed_at FROM state_changes WHERE site_id = ? ORDER BY changed_at DESC LIMIT ?"), siteID, limit)
-	if err != nil {
-		return nil, err
-	}
-	defer rows.Close()
-	var changes []models.StateChange
-	for rows.Next() {
-		var sc models.StateChange
-		if err := rows.Scan(&sc.ID, &sc.SiteID, &sc.FromStatus, &sc.ToStatus, &sc.ErrorReason, &sc.ChangedAt); err != nil {
-			return changes, err
-		}
-		changes = append(changes, sc)
-	}
-	return changes, rows.Err()
-}
-
-func (s *SQLStore) GetStateChangesSince(ctx context.Context, siteID int, since time.Time) ([]models.StateChange, error) {
-	rows, err := s.db.QueryContext(ctx, s.q("SELECT id, site_id, from_status, to_status, error_reason, changed_at FROM state_changes WHERE site_id = ? AND changed_at >= ? ORDER BY changed_at DESC"), siteID, since)
-	if err != nil {
-		return nil, err
-	}
-	defer rows.Close()
-	var changes []models.StateChange
-	for rows.Next() {
-		var sc models.StateChange
-		if err := rows.Scan(&sc.ID, &sc.SiteID, &sc.FromStatus, &sc.ToStatus, &sc.ErrorReason, &sc.ChangedAt); err != nil {
-			return changes, err
-		}
-		changes = append(changes, sc)
-	}
-	return changes, rows.Err()
-}
-
-func (s *SQLStore) SaveCheck(ctx context.Context, siteID int, latencyNs int64, isUp bool) error {
-	return s.SaveCheckFromNode(ctx, siteID, "", latencyNs, isUp)
-}
-
-// SaveCheckFromNode inserts a single check row. Retention is handled out of
-// band by PruneCheckHistory on a timer, not per-insert, to keep the write hot
-// path a plain INSERT.
-func (s *SQLStore) SaveCheckFromNode(ctx context.Context, siteID int, nodeID string, latencyNs int64, isUp bool) error {
-	_, err := s.db.ExecContext(ctx, s.q("INSERT INTO check_history (site_id, node_id, latency_ns, is_up) VALUES (?, ?, ?, ?)"), siteID, nodeID, latencyNs, isUp)
-	return err
-}
-
-// PruneCheckHistory trims check_history to the newest maxCheckHistory rows per
-// site, across all sites, in one pass. Intended to run periodically.
-func (s *SQLStore) PruneCheckHistory(ctx context.Context) error {
-	q := fmt.Sprintf(`DELETE FROM check_history WHERE id IN (
-		SELECT id FROM (
-			SELECT id, ROW_NUMBER() OVER (PARTITION BY site_id ORDER BY checked_at DESC, id DESC) AS rn
-			FROM check_history
-		) ranked WHERE rn > %d
-	)`, maxCheckHistory)
-	_, err := s.db.ExecContext(ctx, s.q(q))
-	return err
-}
-
-// PruneStateChanges trims state_changes to the newest maxStateChangesPerSite
-// rows per site. Generous so realistic SLA windows are unaffected; bounds the
-// otherwise unbounded growth of a flapping monitor's history.
-func (s *SQLStore) PruneStateChanges(ctx context.Context) error {
-	q := fmt.Sprintf(`DELETE FROM state_changes WHERE id IN (
-		SELECT id FROM (
-			SELECT id, ROW_NUMBER() OVER (PARTITION BY site_id ORDER BY changed_at DESC, id DESC) AS rn
-			FROM state_changes
-		) ranked WHERE rn > %d
-	)`, maxStateChangesPerSite)
-	_, err := s.db.ExecContext(ctx, s.q(q))
-	return err
-}
-
-func (s *SQLStore) LoadAllHistory(ctx context.Context, limit int) (map[int][]models.CheckRecord, error) {
-	result := make(map[int][]models.CheckRecord)
-	rows, err := s.db.QueryContext(ctx, s.q(`
-		SELECT site_id, latency_ns, is_up FROM (
-			SELECT site_id, latency_ns, is_up,
-				ROW_NUMBER() OVER (PARTITION BY site_id ORDER BY checked_at DESC) AS rn
-			FROM check_history
-		) sub WHERE rn <= ?`), limit)
-	if err != nil {
-		return result, err
-	}
-	defer rows.Close()
-	for rows.Next() {
-		var r models.CheckRecord
-		if err := rows.Scan(&r.SiteID, &r.LatencyNs, &r.IsUp); err != nil {
-			return result, err
-		}
-		result[r.SiteID] = append(result[r.SiteID], r)
-	}
-	for id, records := range result {
-		for i, j := 0, len(records)-1; i < j; i, j = i+1, j-1 {
-			records[i], records[j] = records[j], records[i]
-		}
-		result[id] = records
-	}
-	return result, rows.Err()
-}

internal/store/sqlstore_history_test.go

@@ -1,134 +0,0 @@
-package store
-
-import (
-	"context"
-	"testing"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-func TestCheckHistory(t *testing.T) {
-	s := newTestStore(t)
-
-	if err := s.SaveCheck(context.Background(), 1, 5000000, true); err != nil {
-		t.Fatalf("SaveCheck: %v", err)
-	}
-	if err := s.SaveCheck(context.Background(), 1, 10000000, false); err != nil {
-		t.Fatalf("SaveCheck: %v", err)
-	}
-	if err := s.SaveCheck(context.Background(), 2, 3000000, true); err != nil {
-		t.Fatalf("SaveCheck site 2: %v", err)
-	}
-
-	history, err := s.LoadAllHistory(context.Background(), 10)
-	if err != nil {
-		t.Fatalf("LoadAllHistory: %v", err)
-	}
-	if len(history[1]) != 2 {
-		t.Fatalf("expected 2 records for site 1, got %d", len(history[1]))
-	}
-	if len(history[2]) != 1 {
-		t.Fatalf("expected 1 record for site 2, got %d", len(history[2]))
-	}
-
-	upCount := 0
-	for _, r := range history[1] {
-		if r.IsUp {
-			upCount++
-		}
-	}
-	if upCount != 1 {
-		t.Errorf("expected 1 up record for site 1, got %d", upCount)
-	}
-}
-
-func TestPruneCheckHistory(t *testing.T) {
-	s := newTestStore(t)
-
-	for i := 0; i < maxCheckHistory+5; i++ {
-		if err := s.SaveCheck(context.Background(), 1, int64(i), true); err != nil {
-			t.Fatalf("SaveCheck site 1: %v", err)
-		}
-	}
-	for i := 0; i < 3; i++ {
-		if err := s.SaveCheck(context.Background(), 2, int64(i), true); err != nil {
-			t.Fatalf("SaveCheck site 2: %v", err)
-		}
-	}
-
-	if err := s.PruneCheckHistory(context.Background()); err != nil {
-		t.Fatalf("PruneCheckHistory: %v", err)
-	}
-
-	history, err := s.LoadAllHistory(context.Background(), maxCheckHistory*2)
-	if err != nil {
-		t.Fatalf("LoadAllHistory: %v", err)
-	}
-	if len(history[1]) != maxCheckHistory {
-		t.Errorf("site 1: expected %d rows after prune, got %d", maxCheckHistory, len(history[1]))
-	}
-	if len(history[2]) != 3 {
-		t.Errorf("site 2: expected 3 rows untouched, got %d", len(history[2]))
-	}
-}
-
-func TestDeleteSiteCascade(t *testing.T) {
-	s := newTestStore(t)
-
-	site := models.SiteConfig{Name: "Cascade Test", URL: "https://example.com", Interval: 30}
-	if err := s.AddSite(context.Background(), site); err != nil {
-		t.Fatalf("AddSite: %v", err)
-	}
-	sites, err := s.GetSites(context.Background())
-	if err != nil {
-		t.Fatalf("GetSites: %v", err)
-	}
-	siteID := sites[0].ID
-
-	if err := s.SaveCheck(context.Background(), siteID, 1000, true); err != nil {
-		t.Fatalf("SaveCheck: %v", err)
-	}
-	if err := s.SaveStateChange(context.Background(), siteID, "UP", "DOWN", "timeout"); err != nil {
-		t.Fatalf("SaveStateChange: %v", err)
-	}
-	mw := models.MaintenanceWindow{
-		MonitorID: siteID,
-		Title:     "Test MW",
-		Type:      "maintenance",
-		StartTime: time.Now(),
-	}
-	if err := s.AddMaintenanceWindow(context.Background(), mw); err != nil {
-		t.Fatalf("AddMaintenanceWindow: %v", err)
-	}
-
-	if err := s.DeleteSite(context.Background(), siteID); err != nil {
-		t.Fatalf("DeleteSite: %v", err)
-	}
-
-	history, err := s.LoadAllHistory(context.Background(), 100)
-	if err != nil {
-		t.Fatalf("LoadAllHistory: %v", err)
-	}
-	if len(history[siteID]) != 0 {
-		t.Errorf("expected 0 check_history rows, got %d", len(history[siteID]))
-	}
-
-	changes, err := s.GetStateChanges(context.Background(), siteID, 100)
-	if err != nil {
-		t.Fatalf("GetStateChanges: %v", err)
-	}
-	if len(changes) != 0 {
-		t.Errorf("expected 0 state_changes rows, got %d", len(changes))
-	}
-
-	windows, err := s.GetActiveMaintenanceWindows(context.Background())
-	if err != nil {
-		t.Fatalf("GetActiveMaintenanceWindows: %v", err)
-	}
-	for _, w := range windows {
-		if w.MonitorID == siteID {
-			t.Errorf("orphaned maintenance window found: id=%d", w.ID)
-		}
-	}
-}

internal/store/sqlstore_maintenance.go

@@ -1,148 +0,0 @@
-package store
-
-import (
-	"context"
-	"database/sql"
-	"fmt"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-func (s *SQLStore) scanMaintenanceWindow(rows *sql.Rows) (models.MaintenanceWindow, error) {
-	var mw models.MaintenanceWindow
-	var endTime sql.NullTime
-	if err := rows.Scan(&mw.ID, &mw.MonitorID, &mw.Title, &mw.Description, &mw.Type, &mw.StartTime, &endTime, &mw.CreatedBy, &mw.CreatedAt); err != nil {
-		return mw, err
-	}
-	if endTime.Valid {
-		mw.EndTime = endTime.Time
-	}
-	return mw, nil
-}
-
-func (s *SQLStore) GetActiveMaintenanceWindows(ctx context.Context) ([]models.MaintenanceWindow, error) {
-	rows, err := s.db.QueryContext(ctx, s.q("SELECT id, monitor_id, title, description, type, start_time, end_time, created_by, created_at FROM maintenance_windows WHERE start_time <= CURRENT_TIMESTAMP AND (end_time IS NULL OR end_time > CURRENT_TIMESTAMP) ORDER BY start_time DESC"))
-	if err != nil {
-		return nil, err
-	}
-	defer rows.Close()
-	var windows []models.MaintenanceWindow
-	for rows.Next() {
-		mw, err := s.scanMaintenanceWindow(rows)
-		if err != nil {
-			return windows, err
-		}
-		windows = append(windows, mw)
-	}
-	return windows, rows.Err()
-}
-
-func (s *SQLStore) GetAllMaintenanceWindows(ctx context.Context, limit int) ([]models.MaintenanceWindow, error) {
-	rows, err := s.db.QueryContext(ctx, s.q("SELECT id, monitor_id, title, description, type, start_time, end_time, created_by, created_at FROM maintenance_windows ORDER BY created_at DESC LIMIT ?"), limit)
-	if err != nil {
-		return nil, err
-	}
-	defer rows.Close()
-	var windows []models.MaintenanceWindow
-	for rows.Next() {
-		mw, err := s.scanMaintenanceWindow(rows)
-		if err != nil {
-			return windows, err
-		}
-		windows = append(windows, mw)
-	}
-	return windows, rows.Err()
-}
-
-func (s *SQLStore) GetOverlappingMaintenanceWindows(ctx context.Context, monitorID int, startTime, endTime time.Time) ([]models.MaintenanceWindow, error) {
-	var timeClause string
-	var args []interface{}
-
-	if endTime.IsZero() {
-		timeClause = "(end_time IS NULL OR end_time > ?)"
-		args = append(args, startTime)
-	} else {
-		timeClause = "(end_time IS NULL OR end_time > ?) AND start_time < ?"
-		args = append(args, startTime, endTime)
-	}
-
-	var scopeClause string
-	if monitorID == 0 {
-		scopeClause = "1=1"
-	} else {
-		scopeClause = "(monitor_id = ? OR monitor_id = 0 OR monitor_id IN (SELECT parent_id FROM sites WHERE id = ? AND parent_id > 0))"
-		args = append(args, monitorID, monitorID)
-	}
-
-	query := fmt.Sprintf(
-		"SELECT id, monitor_id, title, description, type, start_time, end_time, created_by, created_at FROM maintenance_windows WHERE %s AND %s ORDER BY start_time",
-		timeClause, scopeClause,
-	)
-
-	rows, err := s.db.QueryContext(ctx, s.q(query), args...)
-	if err != nil {
-		return nil, err
-	}
-	defer rows.Close()
-
-	var windows []models.MaintenanceWindow
-	for rows.Next() {
-		mw, err := s.scanMaintenanceWindow(rows)
-		if err != nil {
-			return nil, err
-		}
-		windows = append(windows, mw)
-	}
-	return windows, rows.Err()
-}
-
-func (s *SQLStore) AddMaintenanceWindow(ctx context.Context, mw models.MaintenanceWindow) error {
-	if mw.StartTime.IsZero() {
-		mw.StartTime = time.Now()
-	}
-	_, err := s.db.ExecContext(ctx, s.q("INSERT INTO maintenance_windows (monitor_id, title, description, type, start_time, end_time, created_by) VALUES (?, ?, ?, ?, ?, ?, ?)"),
-		mw.MonitorID, mw.Title, mw.Description, mw.Type, mw.StartTime, sql.NullTime{Time: mw.EndTime, Valid: !mw.EndTime.IsZero()}, mw.CreatedBy)
-	return err
-}
-
-func (s *SQLStore) EndMaintenanceWindow(ctx context.Context, id int) error {
-	_, err := s.db.ExecContext(ctx, s.q("UPDATE maintenance_windows SET end_time = CURRENT_TIMESTAMP WHERE id = ?"), id)
-	return err
-}
-
-func (s *SQLStore) DeleteMaintenanceWindow(ctx context.Context, id int) error {
-	_, err := s.db.ExecContext(ctx, s.q("DELETE FROM maintenance_windows WHERE id = ?"), id)
-	if err != nil {
-		return err
-	}
-	s.dialect.ResetSequenceOnEmpty(s.db, "maintenance_windows")
-	return nil
-}
-
-func (s *SQLStore) PruneExpiredMaintenanceWindows(ctx context.Context, retention time.Duration) (int64, error) {
-	cutoff := time.Now().Add(-retention)
-	result, err := s.db.ExecContext(ctx,
-		s.q("DELETE FROM maintenance_windows WHERE end_time IS NOT NULL AND end_time < ?"),
-		cutoff,
-	)
-	if err != nil {
-		return 0, err
-	}
-	return result.RowsAffected()
-}
-
-func (s *SQLStore) IsMonitorInMaintenance(ctx context.Context, monitorID int) (bool, error) {
-	var count int
-	err := s.db.QueryRowContext(ctx, s.q(`SELECT COUNT(*) FROM maintenance_windows
-		WHERE type = 'maintenance'
-		AND start_time <= CURRENT_TIMESTAMP
-		AND (end_time IS NULL OR end_time > CURRENT_TIMESTAMP)
-		AND (monitor_id = 0 OR monitor_id = ?
-			OR monitor_id IN (SELECT parent_id FROM sites WHERE id = ? AND parent_id > 0))`),
-		monitorID, monitorID).Scan(&count)
-	if err != nil {
-		return false, err
-	}
-	return count > 0, nil
-}

internal/store/sqlstore_maintenance_test.go

@@ -1,168 +0,0 @@
-package store
-
-import (
-	"context"
-	"testing"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-)
-
-func TestPruneExpiredMaintenanceWindows(t *testing.T) {
-	s := newTestStore(t)
-
-	now := time.Now()
-
-	// Expired 10 days ago — should be pruned with 7d retention.
-	old := models.MaintenanceWindow{
-		MonitorID: 0,
-		Title:     "Old Window",
-		Type:      "maintenance",
-		StartTime: now.Add(-11 * 24 * time.Hour),
-		EndTime:   now.Add(-10 * 24 * time.Hour),
-	}
-	if err := s.AddMaintenanceWindow(context.Background(), old); err != nil {
-		t.Fatalf("AddMaintenanceWindow (old): %v", err)
-	}
-
-	// Expired 1 day ago — within 7d retention, should survive.
-	recent := models.MaintenanceWindow{
-		MonitorID: 0,
-		Title:     "Recent Window",
-		Type:      "maintenance",
-		StartTime: now.Add(-2 * 24 * time.Hour),
-		EndTime:   now.Add(-1 * 24 * time.Hour),
-	}
-	if err := s.AddMaintenanceWindow(context.Background(), recent); err != nil {
-		t.Fatalf("AddMaintenanceWindow (recent): %v", err)
-	}
-
-	// Ongoing — no end time, should survive.
-	ongoing := models.MaintenanceWindow{
-		MonitorID: 0,
-		Title:     "Ongoing Window",
-		Type:      "maintenance",
-		StartTime: now.Add(-1 * time.Hour),
-	}
-	if err := s.AddMaintenanceWindow(context.Background(), ongoing); err != nil {
-		t.Fatalf("AddMaintenanceWindow (ongoing): %v", err)
-	}
-
-	pruned, err := s.PruneExpiredMaintenanceWindows(context.Background(), 7*24*time.Hour)
-	if err != nil {
-		t.Fatalf("PruneExpiredMaintenanceWindows: %v", err)
-	}
-	if pruned != 1 {
-		t.Errorf("expected 1 pruned, got %d", pruned)
-	}
-
-	all, err := s.GetAllMaintenanceWindows(context.Background(), 100)
-	if err != nil {
-		t.Fatalf("GetAllMaintenanceWindows: %v", err)
-	}
-	if len(all) != 2 {
-		t.Fatalf("expected 2 remaining windows, got %d", len(all))
-	}
-	for _, w := range all {
-		if w.Title == "Old Window" {
-			t.Error("old window should have been pruned")
-		}
-	}
-}
-
-func TestGetOverlappingMaintenanceWindows(t *testing.T) {
-	s := newTestStore(t)
-	ctx := context.Background()
-
-	site := models.SiteConfig{Name: "web", URL: "https://example.com", Interval: 30}
-	if err := s.AddSite(ctx, site); err != nil {
-		t.Fatalf("AddSite: %v", err)
-	}
-
-	now := time.Now()
-
-	active := models.MaintenanceWindow{
-		MonitorID: 1,
-		Title:     "Deploy v2",
-		Type:      "maintenance",
-		StartTime: now.Add(-30 * time.Minute),
-		EndTime:   now.Add(30 * time.Minute),
-	}
-	if err := s.AddMaintenanceWindow(ctx, active); err != nil {
-		t.Fatalf("AddMaintenanceWindow: %v", err)
-	}
-
-	ended := models.MaintenanceWindow{
-		MonitorID: 1,
-		Title:     "Old deploy",
-		Type:      "maintenance",
-		StartTime: now.Add(-3 * time.Hour),
-		EndTime:   now.Add(-2 * time.Hour),
-	}
-	if err := s.AddMaintenanceWindow(ctx, ended); err != nil {
-		t.Fatalf("AddMaintenanceWindow: %v", err)
-	}
-
-	t.Run("same monitor overlaps", func(t *testing.T) {
-		overlaps, err := s.GetOverlappingMaintenanceWindows(ctx, 1, now, now.Add(1*time.Hour))
-		if err != nil {
-			t.Fatalf("unexpected error: %v", err)
-		}
-		if len(overlaps) != 1 {
-			t.Fatalf("expected 1 overlap, got %d", len(overlaps))
-		}
-		if overlaps[0].Title != "Deploy v2" {
-			t.Errorf("expected 'Deploy v2', got %q", overlaps[0].Title)
-		}
-	})
-
-	t.Run("different monitor no overlap", func(t *testing.T) {
-		overlaps, err := s.GetOverlappingMaintenanceWindows(ctx, 99, now, now.Add(1*time.Hour))
-		if err != nil {
-			t.Fatalf("unexpected error: %v", err)
-		}
-		if len(overlaps) != 0 {
-			t.Errorf("expected 0 overlaps, got %d", len(overlaps))
-		}
-	})
-
-	t.Run("global window overlaps all", func(t *testing.T) {
-		global := models.MaintenanceWindow{
-			MonitorID: 0,
-			Title:     "Global freeze",
-			Type:      "maintenance",
-			StartTime: now.Add(-10 * time.Minute),
-			EndTime:   now.Add(2 * time.Hour),
-		}
-		if err := s.AddMaintenanceWindow(ctx, global); err != nil {
-			t.Fatalf("AddMaintenanceWindow: %v", err)
-		}
-		overlaps, err := s.GetOverlappingMaintenanceWindows(ctx, 1, now, now.Add(1*time.Hour))
-		if err != nil {
-			t.Fatalf("unexpected error: %v", err)
-		}
-		if len(overlaps) != 2 {
-			t.Errorf("expected 2 overlaps (specific + global), got %d", len(overlaps))
-		}
-	})
-
-	t.Run("indefinite window overlaps", func(t *testing.T) {
-		overlaps, err := s.GetOverlappingMaintenanceWindows(ctx, 1, now, time.Time{})
-		if err != nil {
-			t.Fatalf("unexpected error: %v", err)
-		}
-		if len(overlaps) < 1 {
-			t.Error("expected at least 1 overlap for indefinite window")
-		}
-	})
-
-	t.Run("ended window excluded", func(t *testing.T) {
-		overlaps, err := s.GetOverlappingMaintenanceWindows(ctx, 1, now.Add(-4*time.Hour), now.Add(-3*time.Hour))
-		if err != nil {
-			t.Fatalf("unexpected error: %v", err)
-		}
-		if len(overlaps) != 0 {
-			t.Errorf("expected 0 overlaps for past range, got %d", len(overlaps))
-		}
-	})
-}

internal/store/sqlstore_test.go

@@ -3,7 +3,9 @@ package store
 import (
 	"context"
 	"fmt"
+	"strings"
 	"testing"
+	"time"
 
 	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
 )
@@ -72,6 +74,60 @@ func TestSiteCRUD(t *testing.T) {
 	}
 }
 
+func TestAlertCRUD(t *testing.T) {
+	s := newTestStore(t)
+
+	if err := s.AddAlert(context.Background(), "Discord", "discord", map[string]string{"url": "https://example.com/hook"}); err != nil {
+		t.Fatalf("AddAlert: %v", err)
+	}
+
+	alerts, err := s.GetAllAlerts(context.Background())
+	if err != nil {
+		t.Fatalf("GetAllAlerts: %v", err)
+	}
+	if len(alerts) != 1 {
+		t.Fatalf("expected 1 alert, got %d", len(alerts))
+	}
+	if alerts[0].Type != "discord" {
+		t.Errorf("expected type 'discord', got '%s'", alerts[0].Type)
+	}
+	if alerts[0].Settings["url"] != "https://example.com/hook" {
+		t.Errorf("settings url mismatch")
+	}
+
+	a, err := s.GetAlert(context.Background(), alerts[0].ID)
+	if err != nil {
+		t.Fatalf("GetAlert: %v", err)
+	}
+	if a.Name != "Discord" {
+		t.Errorf("expected name 'Discord', got '%s'", a.Name)
+	}
+
+	if err := s.UpdateAlert(context.Background(), a.ID, "Slack", "slack", map[string]string{"url": "https://slack.com/hook"}); err != nil {
+		t.Fatalf("UpdateAlert: %v", err)
+	}
+
+	a, err = s.GetAlert(context.Background(), a.ID)
+	if err != nil {
+		t.Fatalf("GetAlert: %v", err)
+	}
+	if a.Type != "slack" {
+		t.Errorf("expected type 'slack', got '%s'", a.Type)
+	}
+
+	if err := s.DeleteAlert(context.Background(), a.ID); err != nil {
+		t.Fatalf("DeleteAlert: %v", err)
+	}
+
+	alerts, err = s.GetAllAlerts(context.Background())
+	if err != nil {
+		t.Fatalf("GetAllAlerts: %v", err)
+	}
+	if len(alerts) != 0 {
+		t.Fatalf("expected 0 alerts after delete, got %d", len(alerts))
+	}
+}
+
 func TestUserCRUD(t *testing.T) {
 	s := newTestStore(t)
 
@@ -245,6 +301,89 @@ func TestImportData_NilUsersPreservesExisting(t *testing.T) {
 	}
 }
 
+func TestCheckHistory(t *testing.T) {
+	s := newTestStore(t)
+
+	if err := s.SaveCheck(context.Background(), 1, 5000000, true); err != nil {
+		t.Fatalf("SaveCheck: %v", err)
+	}
+	if err := s.SaveCheck(context.Background(), 1, 10000000, false); err != nil {
+		t.Fatalf("SaveCheck: %v", err)
+	}
+	if err := s.SaveCheck(context.Background(), 2, 3000000, true); err != nil {
+		t.Fatalf("SaveCheck site 2: %v", err)
+	}
+
+	history, err := s.LoadAllHistory(context.Background(), 10)
+	if err != nil {
+		t.Fatalf("LoadAllHistory: %v", err)
+	}
+	if len(history[1]) != 2 {
+		t.Fatalf("expected 2 records for site 1, got %d", len(history[1]))
+	}
+	if len(history[2]) != 1 {
+		t.Fatalf("expected 1 record for site 2, got %d", len(history[2]))
+	}
+
+	upCount := 0
+	for _, r := range history[1] {
+		if r.IsUp {
+			upCount++
+		}
+	}
+	if upCount != 1 {
+		t.Errorf("expected 1 up record for site 1, got %d", upCount)
+	}
+}
+
+func TestDeleteSiteCascade(t *testing.T) {
+	s := newTestStore(t)
+
+	site := models.SiteConfig{Name: "Cascade Test", URL: "https://example.com", Interval: 30}
+	if err := s.AddSite(context.Background(), site); err != nil {
+		t.Fatalf("AddSite: %v", err)
+	}
+	sites, _ := s.GetSites(context.Background())
+	siteID := sites[0].ID
+
+	if err := s.SaveCheck(context.Background(), siteID, 1000, true); err != nil {
+		t.Fatalf("SaveCheck: %v", err)
+	}
+	if err := s.SaveStateChange(context.Background(), siteID, "UP", "DOWN", "timeout"); err != nil {
+		t.Fatalf("SaveStateChange: %v", err)
+	}
+	mw := models.MaintenanceWindow{
+		MonitorID: siteID,
+		Title:     "Test MW",
+		Type:      "maintenance",
+		StartTime: time.Now(),
+	}
+	if err := s.AddMaintenanceWindow(context.Background(), mw); err != nil {
+		t.Fatalf("AddMaintenanceWindow: %v", err)
+	}
+
+	if err := s.DeleteSite(context.Background(), siteID); err != nil {
+		t.Fatalf("DeleteSite: %v", err)
+	}
+
+	history, _ := s.LoadAllHistory(context.Background(), 100)
+	if len(history[siteID]) != 0 {
+		t.Errorf("expected 0 check_history rows, got %d", len(history[siteID]))
+	}
+
+	changes, _ := s.GetStateChanges(context.Background(), siteID, 100)
+	if len(changes) != 0 {
+		t.Errorf("expected 0 state_changes rows, got %d", len(changes))
+	}
+
+	windows, _ := s.GetActiveMaintenanceWindows(context.Background())
+	for _, w := range windows {
+		if w.MonitorID == siteID {
+			t.Errorf("orphaned maintenance window found: id=%d", w.ID)
+		}
+	}
+}
+
 func TestPruneLogs(t *testing.T) {
 	s := newTestStore(t)
 
@@ -268,7 +407,7 @@ func TestPruneLogs(t *testing.T) {
 	// LoadLogs ordering ties when rows share a created_at second).
 	present := make(map[string]bool, len(logs))
 	for _, l := range logs {
-		present[l.Message] = true
+		present[l] = true
 	}
 	if !present[fmt.Sprintf("log %d", maxLogRows+50-1)] {
 		t.Error("newest log was pruned")
@@ -277,3 +416,134 @@ func TestPruneLogs(t *testing.T) {
 		t.Error("oldest log survived prune")
 	}
 }
+
+func TestPruneCheckHistory(t *testing.T) {
+	s := newTestStore(t)
+
+	for i := 0; i < maxCheckHistory+5; i++ {
+		if err := s.SaveCheck(context.Background(), 1, int64(i), true); err != nil {
+			t.Fatalf("SaveCheck site 1: %v", err)
+		}
+	}
+	for i := 0; i < 3; i++ {
+		if err := s.SaveCheck(context.Background(), 2, int64(i), true); err != nil {
+			t.Fatalf("SaveCheck site 2: %v", err)
+		}
+	}
+
+	if err := s.PruneCheckHistory(context.Background()); err != nil {
+		t.Fatalf("PruneCheckHistory: %v", err)
+	}
+
+	history, err := s.LoadAllHistory(context.Background(), maxCheckHistory*2)
+	if err != nil {
+		t.Fatalf("LoadAllHistory: %v", err)
+	}
+	if len(history[1]) != maxCheckHistory {
+		t.Errorf("site 1: expected %d rows after prune, got %d", maxCheckHistory, len(history[1]))
+	}
+	if len(history[2]) != 3 {
+		t.Errorf("site 2: expected 3 rows untouched, got %d", len(history[2]))
+	}
+}
+
+func TestPruneExpiredMaintenanceWindows(t *testing.T) {
+	s := newTestStore(t)
+
+	now := time.Now()
+
+	// Expired 10 days ago — should be pruned with 7d retention.
+	old := models.MaintenanceWindow{
+		MonitorID: 0,
+		Title:     "Old Window",
+		Type:      "maintenance",
+		StartTime: now.Add(-11 * 24 * time.Hour),
+		EndTime:   now.Add(-10 * 24 * time.Hour),
+	}
+	if err := s.AddMaintenanceWindow(context.Background(), old); err != nil {
+		t.Fatalf("AddMaintenanceWindow (old): %v", err)
+	}
+
+	// Expired 1 day ago — within 7d retention, should survive.
+	recent := models.MaintenanceWindow{
+		MonitorID: 0,
+		Title:     "Recent Window",
+		Type:      "maintenance",
+		StartTime: now.Add(-2 * 24 * time.Hour),
+		EndTime:   now.Add(-1 * 24 * time.Hour),
+	}
+	if err := s.AddMaintenanceWindow(context.Background(), recent); err != nil {
+		t.Fatalf("AddMaintenanceWindow (recent): %v", err)
+	}
+
+	// Ongoing — no end time, should survive.
+	ongoing := models.MaintenanceWindow{
+		MonitorID: 0,
+		Title:     "Ongoing Window",
+		Type:      "maintenance",
+		StartTime: now.Add(-1 * time.Hour),
+	}
+	if err := s.AddMaintenanceWindow(context.Background(), ongoing); err != nil {
+		t.Fatalf("AddMaintenanceWindow (ongoing): %v", err)
+	}
+
+	pruned, err := s.PruneExpiredMaintenanceWindows(context.Background(), 7*24*time.Hour)
+	if err != nil {
+		t.Fatalf("PruneExpiredMaintenanceWindows: %v", err)
+	}
+	if pruned != 1 {
+		t.Errorf("expected 1 pruned, got %d", pruned)
+	}
+
+	all, err := s.GetAllMaintenanceWindows(context.Background(), 100)
+	if err != nil {
+		t.Fatalf("GetAllMaintenanceWindows: %v", err)
+	}
+	if len(all) != 2 {
+		t.Fatalf("expected 2 remaining windows, got %d", len(all))
+	}
+	for _, w := range all {
+		if w.Title == "Old Window" {
+			t.Error("old window should have been pruned")
+		}
+	}
+}
+
+// ImportData must encrypt alert settings (like AddAlert/UpdateAlert) so a
+// restore with UPTOP_ENCRYPTION_KEY set never lands secrets in plaintext.
+func TestImportData_EncryptsAlertSettings(t *testing.T) {
+	s := newTestStore(t)
+	enc, err := NewEncryptor(strings.Repeat("ab", 32)) // 64 hex chars = 32 bytes
+	if err != nil {
+		t.Fatalf("NewEncryptor: %v", err)
+	}
+	s.SetEncryptor(enc)
+
+	backup := models.Backup{
+		Alerts: []models.AlertConfig{
+			{ID: 1, Name: "tg", Type: "telegram", Settings: map[string]string{"token": "123:SECRET", "chat_id": "42"}},
+		},
+	}
+	if err := s.ImportData(context.Background(), backup); err != nil {
+		t.Fatalf("ImportData: %v", err)
+	}
+
+	var raw string
+	if err := s.db.QueryRow("SELECT settings FROM alerts WHERE id = 1").Scan(&raw); err != nil {
+		t.Fatalf("query settings: %v", err)
+	}
+	if !strings.HasPrefix(raw, encryptedPrefix) {
+		t.Errorf("imported settings not encrypted: %q", raw)
+	}
+	if strings.Contains(raw, "SECRET") {
+		t.Errorf("plaintext secret found in stored column: %q", raw)
+	}
+
+	alerts, err := s.GetAllAlerts(context.Background())
+	if err != nil {
+		t.Fatalf("GetAllAlerts: %v", err)
+	}
+	if len(alerts) != 1 || alerts[0].Settings["token"] != "123:SECRET" {
+		t.Errorf("decrypt round-trip failed: %+v", alerts)
+	}
+}

internal/store/store.go

@@ -61,13 +61,12 @@ type Store interface {
 
 	// Logs
 	SaveLog(ctx context.Context, message string) error
-	LoadLogs(ctx context.Context, limit int) ([]models.LogEntry, error)
+	LoadLogs(ctx context.Context, limit int) ([]string, error)
 	PruneLogs(ctx context.Context) error
 
 	// Maintenance Windows
 	GetActiveMaintenanceWindows(ctx context.Context) ([]models.MaintenanceWindow, error)
 	GetAllMaintenanceWindows(ctx context.Context, limit int) ([]models.MaintenanceWindow, error)
-	GetOverlappingMaintenanceWindows(ctx context.Context, monitorID int, startTime, endTime time.Time) ([]models.MaintenanceWindow, error)
 	AddMaintenanceWindow(ctx context.Context, mw models.MaintenanceWindow) error
 	EndMaintenanceWindow(ctx context.Context, id int) error
 	DeleteMaintenanceWindow(ctx context.Context, id int) error

internal/store/storetest/mock.go

@@ -212,8 +212,8 @@ func (m *BaseMock) SaveLog(ctx context.Context, message string) error {
 	return nil
 }
 
-func (m *BaseMock) LoadLogs(_ context.Context, _ int) ([]models.LogEntry, error) { return nil, nil }
-func (m *BaseMock) PruneLogs(_ context.Context) error                            { return nil }
+func (m *BaseMock) LoadLogs(_ context.Context, _ int) ([]string, error) { return nil, nil }
+func (m *BaseMock) PruneLogs(_ context.Context) error                   { return nil }
 
 func (m *BaseMock) GetActiveMaintenanceWindows(ctx context.Context) ([]models.MaintenanceWindow, error) {
 	if m.GetActiveMaintenanceWindowsFunc != nil {
@@ -229,10 +229,6 @@ func (m *BaseMock) GetAllMaintenanceWindows(ctx context.Context, limit int) ([]m
 	return nil, nil
 }
 
-func (m *BaseMock) GetOverlappingMaintenanceWindows(_ context.Context, _ int, _, _ time.Time) ([]models.MaintenanceWindow, error) {
-	return nil, nil
-}
-
 func (m *BaseMock) AddMaintenanceWindow(_ context.Context, _ models.MaintenanceWindow) error {
 	return nil
 }

internal/tui/chart.go

@@ -1,132 +0,0 @@
-package tui
-
-import (
-	"fmt"
-	"strings"
-	"time"
-)
-
-func (m Model) latencyChart(latencies []time.Duration, statuses []bool, width, height int) string {
-	if len(latencies) == 0 || width < 20 || height < 1 {
-		return ""
-	}
-
-	var minMs, maxMs, sumMs int64
-	minMs = latencies[0].Milliseconds()
-	maxMs = minMs
-	upCount := 0
-	for i, l := range latencies {
-		ms := l.Milliseconds()
-		if i < len(statuses) && !statuses[i] {
-			continue
-		}
-		if ms < minMs {
-			minMs = ms
-		}
-		if ms > maxMs {
-			maxMs = ms
-		}
-		sumMs += ms
-		upCount++
-	}
-	var avgMs int64
-	if upCount > 0 {
-		avgMs = sumMs / int64(upCount)
-	}
-
-	maxLabel := fmt.Sprintf("%dms", maxMs)
-	minLabel := fmt.Sprintf("%dms", minMs)
-	labelW := len(maxLabel)
-	if len(minLabel) > labelW {
-		labelW = len(minLabel)
-	}
-	labelW += 1
-
-	chartW := width - labelW
-	if chartW > len(latencies) {
-		chartW = len(latencies)
-	}
-	if chartW < 10 {
-		chartW = 10
-	}
-
-	samples := latencies
-	sampledStatuses := statuses
-	if len(samples) > chartW {
-		samples = samples[len(samples)-chartW:]
-		if len(sampledStatuses) > chartW {
-			sampledStatuses = sampledStatuses[len(sampledStatuses)-chartW:]
-		}
-	}
-
-	spread := maxMs - minMs
-	if spread == 0 {
-		spread = 1
-	}
-	totalLevels := height * 8
-
-	levels := make([]int, len(samples))
-	for i, l := range samples {
-		ms := l.Milliseconds()
-		if i < len(sampledStatuses) && !sampledStatuses[i] {
-			levels[i] = 0
-			continue
-		}
-		lvl := int(float64(ms-minMs) / float64(spread) * float64(totalLevels-1))
-		if lvl < 1 && ms > 0 {
-			lvl = 1
-		}
-		if lvl >= totalLevels {
-			lvl = totalLevels - 1
-		}
-		levels[i] = lvl
-	}
-
-	labelStyle := m.st.subtleStyle
-
-	var rows []string
-	for row := height - 1; row >= 0; row-- {
-		var label string
-		switch row {
-		case height - 1:
-			label = fmt.Sprintf("%*s", labelW, maxLabel)
-		case 0:
-			label = fmt.Sprintf("%*s", labelW, minLabel)
-		default:
-			label = strings.Repeat(" ", labelW)
-		}
-
-		var sb strings.Builder
-		sb.WriteString(labelStyle.Render(label))
-
-		rowBase := row * 8
-		for i := range samples {
-			fill := levels[i] - rowBase
-			if fill <= 0 {
-				sb.WriteString(" ")
-				continue
-			}
-			if fill > 8 {
-				fill = 8
-			}
-			ch := string(sparkChars[fill-1])
-
-			isDown := i < len(sampledStatuses) && !sampledStatuses[i]
-			if isDown {
-				sb.WriteString(m.st.dangerStyle.Render(ch))
-			} else {
-				sb.WriteString(m.latencyStyle(samples[i].Milliseconds(), nil).Render(ch))
-			}
-		}
-		rows = append(rows, sb.String())
-	}
-
-	stats := fmt.Sprintf("%*s Min %s  Avg %s  Max %s",
-		labelW, "",
-		m.fmtLatency(time.Duration(minMs)*time.Millisecond),
-		m.fmtLatency(time.Duration(avgMs)*time.Millisecond),
-		m.fmtLatency(time.Duration(maxMs)*time.Millisecond))
-	rows = append(rows, stats)
-
-	return strings.Join(rows, "\n")
-}

internal/tui/const.go

@@ -1,20 +0,0 @@
-package tui
-
-import "time"
-
-const (
-	nodeOnlineThreshold = 60 * time.Second
-	nodeStaleThreshold  = 5 * time.Minute
-
-	uptimeGoodPct      = 99.0
-	uptimeExcellentPct = 99.9
-	uptimeWarnPct      = 95.0
-	uptimePrecisionPct = 99.99
-
-	stateHistoryLookback = 30 * 24 * time.Hour
-	stateHistoryDays     = 30
-	stateHistoryLimit    = 100
-
-	httpErrorThreshold = 400
-	errorDetailMaxLen  = 30
-)

internal/tui/data.go

@@ -8,14 +8,13 @@ import (
 	"time"
 
 	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/monitor"
 	"gitea.lerkolabs.com/lerkolabs/uptop/internal/store"
 	tea "github.com/charmbracelet/bubbletea"
 )
 
-func loadCollapsed(ctx context.Context, s store.Store) map[int]bool {
+func loadCollapsed(s store.Store) map[int]bool {
 	m := make(map[int]bool)
-	raw, err := s.GetPreference(ctx, "collapsed_groups")
+	raw, err := s.GetPreference(context.Background(), "collapsed_groups")
 	if err != nil || raw == "" {
 		return m
 	}
@@ -50,22 +49,7 @@ func writeCmd(op string, fn func() error) tea.Cmd {
 	}
 }
 
-func (m *Model) saveBottomPanelPref() tea.Cmd {
-	v := "logs"
-	switch m.bottomPanel {
-	case bottomNone:
-		v = "none"
-	case bottomMaint:
-		v = "maint"
-	}
-	st := m.store
-	ctx := m.ctx
-	return writeCmd("Save bottom panel preference", func() error {
-		return st.SetPreference(ctx, "bottom_panel", v)
-	})
-}
-
-func sortSitesForDisplay(allSites []models.Site, collapsed map[int]bool, sortCol int, sortAsc bool) []models.Site {
+func sortSitesForDisplay(allSites []models.Site, collapsed map[int]bool) []models.Site {
 	var groups, ungrouped []models.Site
 	children := make(map[int][]models.Site)
 	for _, s := range allSites {
@@ -77,32 +61,15 @@ func sortSitesForDisplay(allSites []models.Site, collapsed map[int]bool, sortCol
 			ungrouped = append(ungrouped, s)
 		}
 	}
-	sortSlice := func(s []models.Site) {
-		sort.Slice(s, func(i, j int) bool { return s[i].ID < s[j].ID })
-		sort.SliceStable(s, func(i, j int) bool {
-			var less bool
-			switch sortCol {
-			case sortName:
-				less = strings.ToLower(s[i].Name) < strings.ToLower(s[j].Name)
-			case sortLatency:
-				less = s[i].Latency < s[j].Latency
-			default:
-				less = siteOrder(s[i]) < siteOrder(s[j])
-			}
-			if !sortAsc {
-				return less
-			}
-			return !less
-		})
-	}
-
-	sortSlice(groups)
+	sort.Slice(groups, func(i, j int) bool { return groups[i].ID < groups[j].ID })
 	for pid := range children {
 		c := children[pid]
-		sortSlice(c)
+		sort.Slice(c, func(i, j int) bool { return c[i].ID < c[j].ID })
+		sort.SliceStable(c, func(i, j int) bool { return siteOrder(c[i]) < siteOrder(c[j]) })
 		children[pid] = c
 	}
-	sortSlice(ungrouped)
+	sort.Slice(ungrouped, func(i, j int) bool { return ungrouped[i].ID < ungrouped[j].ID })
+	sort.SliceStable(ungrouped, func(i, j int) bool { return siteOrder(ungrouped[i]) < siteOrder(ungrouped[j]) })
 
 	var ordered []models.Site
 	for _, g := range groups {
@@ -132,15 +99,14 @@ func filterSites(sites []models.Site, needle string) []models.Site {
 // separately via loadTabDataCmd.
 func (m *Model) refreshLive() {
 	allSites := m.engine.GetAllSites()
-	ordered := sortSitesForDisplay(allSites, m.collapsed, m.sortColumn, m.sortAsc)
+	ordered := sortSitesForDisplay(allSites, m.collapsed)
 	if m.filterText != "" {
 		ordered = filterSites(ordered, m.filterText)
 	}
 	m.sites = ordered
-	m.buildMaintSet()
 	m.refreshLogContent()
 
-	if m.selectedID != 0 {
+	if m.currentTab == 0 && m.selectedID != 0 {
 		for i, s := range m.sites {
 			if s.ID == m.selectedID {
 				m.cursor = i
@@ -152,7 +118,7 @@ func (m *Model) refreshLive() {
 }
 
 func (m *Model) syncSelectedID() {
-	if m.cursor < len(m.sites) {
+	if m.currentTab == 0 && m.cursor < len(m.sites) {
 		m.selectedID = m.sites[m.cursor].ID
 	}
 }
@@ -168,118 +134,6 @@ func (m *Model) clampCursor() {
 	}
 }
 
-func (m *Model) detailCmdIfNeeded() tea.Cmd {
-	if m.focusedPanel == panelMonitors && m.detailOpen && m.cursor < len(m.sites) {
-		m.detailMode = detailDefault
-		m.detailScrollOffset = 0
-		return m.loadDetailCmd(m.sites[m.cursor].ID)
-	}
-	return nil
-}
-
-func (m *Model) jumpToTop() {
-	switch m.focusedPanel {
-	case panelDetail:
-		m.detailScrollOffset = 0
-	case panelMaint:
-		m.maintCursor = 0
-	case panelLogs:
-		m.logScrollOffset = 0
-	default:
-		m.cursor = 0
-		m.tableOffset = 0
-		m.syncSelectedID()
-	}
-}
-
-func (m *Model) jumpToBottom() {
-	switch m.focusedPanel {
-	case panelDetail:
-		m.detailScrollOffset = 9999
-	case panelMaint:
-		windows := m.activeMaintWindows()
-		if len(windows) > 0 {
-			m.maintCursor = len(windows) - 1
-		}
-	case panelLogs:
-		total := m.filteredLogCount()
-		if total > 0 {
-			m.logScrollOffset = total - 1
-		}
-	default:
-		max := m.currentListLen() - 1
-		if max >= 0 {
-			m.cursor = max
-			if m.cursor >= m.tableOffset+m.maxTableRows {
-				m.tableOffset = m.cursor - m.maxTableRows + 1
-			}
-			m.syncSelectedID()
-		}
-	}
-}
-
-func (m *Model) halfPageUp() {
-	half := m.maxTableRows / 2
-	if half < 1 {
-		half = 1
-	}
-	switch m.focusedPanel {
-	case panelDetail:
-		m.detailScrollOffset -= half
-		if m.detailScrollOffset < 0 {
-			m.detailScrollOffset = 0
-		}
-	case panelMaint:
-		m.maintCursor -= half
-		if m.maintCursor < 0 {
-			m.maintCursor = 0
-		}
-	case panelLogs:
-		m.scrollLogs(-half)
-	default:
-		m.cursor -= half
-		if m.cursor < 0 {
-			m.cursor = 0
-		}
-		if m.cursor < m.tableOffset {
-			m.tableOffset = m.cursor
-		}
-		m.syncSelectedID()
-	}
-}
-
-func (m *Model) halfPageDown() {
-	half := m.maxTableRows / 2
-	if half < 1 {
-		half = 1
-	}
-	switch m.focusedPanel {
-	case panelDetail:
-		m.detailScrollOffset += half
-	case panelMaint:
-		windows := m.activeMaintWindows()
-		m.maintCursor += half
-		if len(windows) > 0 && m.maintCursor >= len(windows) {
-			m.maintCursor = len(windows) - 1
-		}
-	case panelLogs:
-		m.scrollLogs(half)
-	default:
-		max := m.currentListLen() - 1
-		m.cursor += half
-		if m.cursor > max {
-			m.cursor = max
-		}
-		if m.cursor < 0 {
-			m.cursor = 0
-		}
-		if m.cursor >= m.tableOffset+m.maxTableRows {
-			m.tableOffset = m.cursor - m.maxTableRows + 1
-		}
-		m.syncSelectedID()
-	}
-}
-
 // loadTabDataCmd returns a tea.Cmd that loads the DB-backed tab tables off the
 // UI goroutine. Each call bumps tabSeq and stamps the reply with it, so
 // handleTabData can drop out-of-order results from slower earlier loads. The
@@ -290,9 +144,9 @@ func (m *Model) loadTabDataCmd() tea.Cmd {
 	m.tabSeq++
 	seq := m.tabSeq
 	st := m.store
-	ctx := m.ctx
 	isAdmin := m.isAdmin
 	return func() tea.Msg {
+		ctx := context.Background()
 		alerts, err := st.GetAllAlerts(ctx)
 		if err != nil {
 			return tabDataMsg{seq: seq, err: err}
@@ -319,20 +173,8 @@ func (m *Model) loadTabDataCmd() tea.Cmd {
 // goroutine. View renders the cached result rather than querying the DB.
 func (m *Model) loadDetailCmd(siteID int) tea.Cmd {
 	eng := m.engine
-	ctx := m.ctx
-	var currentStatus models.Status
-	for _, s := range m.sites {
-		if s.ID == siteID {
-			currentStatus = s.Status
-			break
-		}
-	}
 	return func() tea.Msg {
-		changes := eng.GetStateChanges(ctx, siteID, 5)
-		now := time.Now()
-		allChanges := eng.GetStateChangesSince(ctx, siteID, now.Add(-stateHistoryLookback))
-		daily := monitor.ComputeDailyBreakdown(allChanges, currentStatus, stateHistoryDays, now)
-		return detailDataMsg{siteID: siteID, changes: changes, dailyDays: daily}
+		return detailDataMsg{siteID: siteID, changes: eng.GetStateChanges(siteID, 5)}
 	}
 }
 
@@ -340,9 +182,8 @@ func (m *Model) loadDetailCmd(siteID int) tea.Cmd {
 // the UI goroutine.
 func (m *Model) loadHistoryCmd(siteID int) tea.Cmd {
 	eng := m.engine
-	ctx := m.ctx
 	return func() tea.Msg {
-		return historyDataMsg{siteID: siteID, changes: eng.GetStateChanges(ctx, siteID, stateHistoryLimit)}
+		return historyDataMsg{siteID: siteID, changes: eng.GetStateChanges(siteID, 100)}
 	}
 }
 
@@ -351,13 +192,12 @@ func (m *Model) loadHistoryCmd(siteID int) tea.Cmd {
 // can be recognized and dropped.
 func (m *Model) loadSLACmd(siteID, periodIdx int) tea.Cmd {
 	eng := m.engine
-	ctx := m.ctx
 	since := time.Now().Add(-slaPeriods[periodIdx].duration)
 	return func() tea.Msg {
 		return slaDataMsg{
 			siteID:    siteID,
 			periodIdx: periodIdx,
-			changes:   eng.GetStateChangesSince(ctx, siteID, since),
+			changes:   eng.GetStateChangesSince(siteID, since),
 		}
 	}
 }

internal/tui/data_test.go

@@ -12,7 +12,7 @@ func TestSortSitesForDisplay_GroupsFirst(t *testing.T) {
 		{SiteConfig: models.SiteConfig{ID: 1, Name: "group-a", Type: "group"}, SiteState: models.SiteState{Status: "UP"}},
 		{SiteConfig: models.SiteConfig{ID: 2, Name: "child", Type: "http", ParentID: 1}, SiteState: models.SiteState{Status: "UP"}},
 	}
-	result := sortSitesForDisplay(sites, nil, sortStatus, false)
+	result := sortSitesForDisplay(sites, nil)
 	if len(result) != 3 {
 		t.Fatalf("expected 3 sites, got %d", len(result))
 	}
@@ -34,7 +34,7 @@ func TestSortSitesForDisplay_CollapsedHidesChildren(t *testing.T) {
 		{SiteConfig: models.SiteConfig{ID: 3, Name: "child-2", Type: "http", ParentID: 1}, SiteState: models.SiteState{Status: "UP"}},
 	}
 	collapsed := map[int]bool{1: true}
-	result := sortSitesForDisplay(sites, collapsed, sortStatus, false)
+	result := sortSitesForDisplay(sites, collapsed)
 	if len(result) != 1 {
 		t.Fatalf("collapsed group should hide children, got %d items", len(result))
 	}
@@ -49,7 +49,7 @@ func TestSortSitesForDisplay_StatusOrdering(t *testing.T) {
 		{SiteConfig: models.SiteConfig{ID: 2, Name: "down-site", Type: "http"}, SiteState: models.SiteState{Status: "DOWN"}},
 		{SiteConfig: models.SiteConfig{ID: 3, Name: "late-site", Type: "http"}, SiteState: models.SiteState{Status: "LATE"}},
 	}
-	result := sortSitesForDisplay(sites, nil, sortStatus, false)
+	result := sortSitesForDisplay(sites, nil)
 	if result[0].Status != "DOWN" {
 		t.Errorf("DOWN should sort first, got %s", result[0].Status)
 	}

internal/tui/errclass.go

@@ -58,7 +58,7 @@ func classifyError(errorReason string, siteType string, statusCode int) ErrorCat
 	if strings.HasPrefix(lower, "http ") || strings.Contains(lower, "keyword not found") {
 		return ErrCatHTTP
 	}
-	if statusCode >= httpErrorThreshold {
+	if statusCode >= 400 {
 		return ErrCatHTTP
 	}
 
@@ -145,7 +145,7 @@ func extractDetail(errorReason string, cat ErrorCategory, statusCode int) string
 			}
 		}
 		if detail == "" {
-			detail = limitStr(errorReason, errorDetailMaxLen)
+			detail = limitStr(errorReason, 30)
 		}
 	case ErrCatTCP:
 		for _, keyword := range []string{"connection refused", "connection reset", "no route to host", "network unreachable"} {
@@ -155,7 +155,7 @@ func extractDetail(errorReason string, cat ErrorCategory, statusCode int) string
 			}
 		}
 		if detail == "" {
-			detail = limitStr(errorReason, errorDetailMaxLen)
+			detail = limitStr(errorReason, 30)
 		}
 	case ErrCatTLS:
 		for _, keyword := range []string{"certificate expired", "certificate has expired", "handshake failure", "unknown authority"} {
@@ -166,16 +166,16 @@ func extractDetail(errorReason string, cat ErrorCategory, statusCode int) string
 		}
 		if detail == "" && strings.Contains(lower, "x509:") {
 			idx := strings.Index(lower, "x509:")
-			detail = limitStr(errorReason[idx:], errorDetailMaxLen)
+			detail = limitStr(errorReason[idx:], 30)
 		}
 		if detail == "" {
-			detail = limitStr(errorReason, errorDetailMaxLen)
+			detail = limitStr(errorReason, 30)
 		}
 	case ErrCatHTTP:
 		if statusCode > 0 {
 			detail = fmt.Sprintf("HTTP %d", statusCode)
 		} else {
-			detail = limitStr(errorReason, errorDetailMaxLen)
+			detail = limitStr(errorReason, 30)
 		}
 	case ErrCatTimeout:
 		if strings.Contains(lower, "i/o timeout") {
@@ -187,12 +187,12 @@ func extractDetail(errorReason string, cat ErrorCategory, statusCode int) string
 		if strings.Contains(lower, "no icmp response") {
 			detail = "no response"
 		} else {
-			detail = limitStr(errorReason, errorDetailMaxLen)
+			detail = limitStr(errorReason, 30)
 		}
 	case ErrCatPrivate:
 		detail = "private IP blocked"
 	default:
-		detail = limitStr(errorReason, errorDetailMaxLen)
+		detail = limitStr(errorReason, 30)
 	}
 
 	return detail

internal/tui/format.go

@@ -49,13 +49,13 @@ func siteOrder(s models.Site) int {
 		return 3
 	}
 	switch s.Status {
-	case models.StatusDown, models.StatusSSLExp:
+	case "DOWN", "SSL EXP":
 		return 0
-	case models.StatusStale:
+	case "STALE":
 		return 1
-	case models.StatusLate:
+	case "LATE":
 		return 1
-	case models.StatusPending:
+	case "PENDING":
 		return 3
 	default:
 		return 2
@@ -104,13 +104,6 @@ func (m Model) fmtLatency(d time.Duration) string {
 	return m.st.dangerStyle.Render(s)
 }
 
-func (m Model) fmtUptimeMaint(statuses []bool, siteID int) string {
-	if m.isMonitorInMaintenance(siteID) {
-		return m.st.subtleStyle.Render("—")
-	}
-	return m.fmtUptime(statuses)
-}
-
 func (m Model) fmtUptime(statuses []bool) string {
 	if len(statuses) == 0 {
 		return m.st.subtleStyle.Render("—")
@@ -162,27 +155,6 @@ func (m Model) fmtRetries(site models.Site) string {
 	return s
 }
 
-func (m Model) fmtStatusDot(status models.Status, paused bool, inMaint bool) string {
-	if paused {
-		return m.st.warnStyle.Render("◇")
-	}
-	if inMaint {
-		return m.st.maintStyle.Render("◼")
-	}
-	switch status {
-	case models.StatusDown, models.StatusSSLExp:
-		return m.st.dangerStyle.Render("▼")
-	case models.StatusLate:
-		return m.st.warnStyle.Render("◆")
-	case models.StatusStale:
-		return m.st.staleStyle.Render("◆")
-	case models.StatusPending:
-		return m.st.subtleStyle.Render("○")
-	default:
-		return m.st.specialStyle.Render("▲")
-	}
-}
-
 func (m Model) fmtStatus(status models.Status, paused bool, inMaint bool) string {
 	if paused {
 		return m.st.warnStyle.Render("◇ PAUSED")

internal/tui/histogram.go

@@ -1,89 +0,0 @@
-package tui
-
-import (
-	"fmt"
-	"strings"
-	"time"
-
-	"github.com/charmbracelet/lipgloss"
-)
-
-type latencyBucket struct {
-	label string
-	min   int64
-	max   int64
-	count int
-}
-
-func (m Model) latencyHistogram(latencies []time.Duration, statuses []bool, width int) string {
-	if len(latencies) == 0 || width < 30 {
-		return ""
-	}
-
-	buckets := []latencyBucket{
-		{"0-50ms", 0, 50, 0},
-		{"50-100ms", 50, 100, 0},
-		{"100-200ms", 100, 200, 0},
-		{"200-500ms", 200, 500, 0},
-		{"500ms+", 500, 999999, 0},
-	}
-
-	for i, l := range latencies {
-		if i < len(statuses) && !statuses[i] {
-			continue
-		}
-		ms := l.Milliseconds()
-		for j := range buckets {
-			if ms >= buckets[j].min && ms < buckets[j].max {
-				buckets[j].count++
-				break
-			}
-		}
-	}
-
-	maxCount := 0
-	for _, b := range buckets {
-		if b.count > maxCount {
-			maxCount = b.count
-		}
-	}
-	if maxCount == 0 {
-		return ""
-	}
-
-	labelW := 10
-	countW := len(fmt.Sprintf("%d", maxCount))
-	barW := width - labelW - countW - 4
-	if barW < 5 {
-		barW = 5
-	}
-
-	var sb strings.Builder
-	for _, b := range buckets {
-		fill := 0
-		if maxCount > 0 {
-			fill = b.count * barW / maxCount
-		}
-
-		var barColor lipgloss.Style
-		switch {
-		case b.min < 100:
-			barColor = m.st.specialStyle
-		case b.min < 500:
-			barColor = m.st.warnStyle
-		default:
-			barColor = m.st.dangerStyle
-		}
-
-		bar := barColor.Render(strings.Repeat("█", fill))
-		if fill < barW {
-			bar += m.st.subtleStyle.Render(strings.Repeat("░", barW-fill))
-		}
-
-		label := fmt.Sprintf("%*s", labelW, b.label)
-		count := fmt.Sprintf("%*d", countW, b.count)
-		sb.WriteString("  " + m.st.subtleStyle.Render(label) + " " + bar + " " + count + "\n")
-	}
-
-	return sb.String()
-}

internal/tui/messages.go

@@ -4,7 +4,6 @@ import (
 	"time"
 
 	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/monitor"
 )
 
 // tabRefreshTTL bounds how often the DB-backed tab data (alerts, users, nodes,
@@ -35,9 +34,8 @@ type tabDataMsg struct {
 // on entry and refreshed on the tab-data cadence so View never touches the
 // database.
 type detailDataMsg struct {
-	siteID    int
-	changes   []models.StateChange
-	dailyDays []monitor.DayReport
+	siteID  int
+	changes []models.StateChange
 }
 
 // historyDataMsg carries the full state-change history for the history view.

internal/tui/panel.go

@@ -1,181 +0,0 @@
-package tui
-
-import (
-	"strings"
-
-	"github.com/charmbracelet/lipgloss"
-)
-
-type scrollbar struct {
-	pos     int
-	total   int
-	visible int
-}
-
-func (m Model) titledPanelH(title, content, footer string, width, height, scrollOffset int, sb scrollbar, focused bool) string {
-	if height <= 0 {
-		return m.titledPanel(title, content, width, focused)
-	}
-
-	borderColor := m.theme.Border
-	titleColor := m.theme.Muted
-	if focused {
-		borderColor = m.theme.Accent
-		titleColor = m.theme.Accent
-	}
-
-	bc := lipgloss.NewStyle().Foreground(borderColor)
-	tc := lipgloss.NewStyle().Foreground(titleColor).Bold(true)
-
-	innerW := width - 2
-	if innerW < 10 {
-		innerW = 10
-	}
-
-	titleRendered := tc.Render(" " + title + " ")
-	titleLen := len([]rune(title)) + 2
-	fillLen := innerW - titleLen - 1
-	if fillLen < 0 {
-		fillLen = 0
-	}
-
-	top := bc.Render("╭─") + titleRendered + bc.Render(strings.Repeat("─", fillLen)+"╮")
-	bottom := bc.Render("╰" + strings.Repeat("─", innerW) + "╯")
-
-	contentStyle := lipgloss.NewStyle().Width(innerW).MaxWidth(innerW)
-	inner := contentStyle.Render(content)
-	contentLines := strings.Split(inner, "\n")
-
-	var footerLines []string
-	if footer != "" {
-		footerRendered := contentStyle.Render(footer)
-		footerLines = strings.Split(footerRendered, "\n")
-	}
-
-	bodyH := height - 2 - len(footerLines)
-	if bodyH < 1 {
-		bodyH = 1
-	}
-
-	if scrollOffset > len(contentLines)-bodyH {
-		scrollOffset = len(contentLines) - bodyH
-	}
-	if scrollOffset < 0 {
-		scrollOffset = 0
-	}
-
-	end := scrollOffset + bodyH
-	if end > len(contentLines) {
-		end = len(contentLines)
-	}
-	visible := contentLines[scrollOffset:end]
-
-	if sb.total == 0 && len(contentLines) > bodyH {
-		sb = scrollbar{pos: scrollOffset, total: len(contentLines), visible: bodyH}
-	}
-	sbVisible := sb.visible
-	if sbVisible <= 0 {
-		sbVisible = bodyH
-	}
-	showScrollbar := sb.total > 0 && sb.total > sbVisible
-	var thumbStart, thumbEnd int
-	if showScrollbar {
-		thumbSize := bodyH * sbVisible / sb.total
-		if thumbSize < 1 {
-			thumbSize = 1
-		}
-		scrollRange := sb.total - sbVisible
-		if scrollRange < 1 {
-			scrollRange = 1
-		}
-		trackSpace := bodyH - thumbSize
-		thumbStart = sb.pos * trackSpace / scrollRange
-		if thumbStart < 0 {
-			thumbStart = 0
-		}
-		thumbEnd = thumbStart + thumbSize
-		if thumbEnd > bodyH {
-			thumbEnd = bodyH
-		}
-	}
-
-	scrollTrack := lipgloss.NewStyle().Foreground(m.theme.Border).Render("░")
-	scrollThumb := lipgloss.NewStyle().Foreground(m.theme.Accent).Render("█")
-
-	borderLine := func(line string, idx int) string {
-		rightBorder := bc.Render("│")
-		if showScrollbar && idx >= thumbStart && idx < thumbEnd {
-			rightBorder = scrollThumb
-		} else if showScrollbar {
-			rightBorder = scrollTrack
-		}
-		return bc.Render("│") + line + strings.Repeat(" ", max(0, innerW-lipgloss.Width(line))) + rightBorder
-	}
-	emptyLine := func(idx int) string {
-		return borderLine(strings.Repeat(" ", innerW), idx)
-	}
-
-	var lines []string
-	lines = append(lines, top)
-	for i, line := range visible {
-		lines = append(lines, borderLine(line, i))
-	}
-	for i := len(visible); len(lines) < height-1-len(footerLines); i++ {
-		lines = append(lines, emptyLine(i))
-	}
-	for _, line := range footerLines {
-		lines = append(lines, borderLine(line, -1))
-	}
-	lines = append(lines, bottom)
-
-	return strings.Join(lines, "\n")
-}
-
-func (m Model) titledPanel(title, content string, width int, focused bool) string {
-	borderColor := m.theme.Border
-	titleColor := m.theme.Muted
-	if focused {
-		borderColor = m.theme.Accent
-		titleColor = m.theme.Accent
-	}
-
-	bc := lipgloss.NewStyle().Foreground(borderColor)
-	tc := lipgloss.NewStyle().Foreground(titleColor).Bold(true)
-
-	innerW := width - 2
-	if innerW < 10 {
-		innerW = 10
-	}
-
-	titleRendered := tc.Render(" " + title + " ")
-	titleLen := len([]rune(title)) + 2
-	fillLen := innerW - titleLen - 1
-	if fillLen < 0 {
-		fillLen = 0
-	}
-
-	top := bc.Render("╭─") + titleRendered + bc.Render(strings.Repeat("─", fillLen)+"╮")
-
-	contentStyle := lipgloss.NewStyle().Width(innerW).MaxWidth(innerW)
-	inner := contentStyle.Render(content)
-
-	var lines []string
-	lines = append(lines, top)
-	for _, line := range strings.Split(inner, "\n") {
-		lines = append(lines, bc.Render("│")+line+strings.Repeat(" ", max(0, innerW-lipgloss.Width(line)))+bc.Render("│"))
-	}
-	lines = append(lines, bc.Render("╰"+strings.Repeat("─", innerW)+"╯"))
-
-	return strings.Join(lines, "\n")
-}
-
-func max(a, b int) int {
-	if a > b {
-		return a
-	}
-	return b
-}
-
-func placeOverlay(fg string, termW, termH int) string {
-	return lipgloss.Place(termW, termH, lipgloss.Center, lipgloss.Center, fg)
-}

internal/tui/sparkline.go

@@ -17,17 +17,6 @@ func parseHex(hex string) (r, g, b uint8) {
 	return
 }
 
-func trueColorHex(c lipgloss.TerminalColor) string {
-	switch v := c.(type) {
-	case lipgloss.CompleteColor:
-		return v.TrueColor
-	case lipgloss.Color:
-		return string(v)
-	default:
-		return ""
-	}
-}
-
 func dimColor(hex string, brightness float64) lipgloss.Color {
 	r, g, b := parseHex(hex)
 	f := 0.3 + brightness*0.7
@@ -38,36 +27,35 @@ func dimColor(hex string, brightness float64) lipgloss.Color {
 	))
 }
 
-func withBg(s lipgloss.Style, bg lipgloss.TerminalColor) lipgloss.Style {
-	if bg != nil {
+func withBg(s lipgloss.Style, bg lipgloss.Color) lipgloss.Style {
+	if bg != "" {
 		return s.Background(bg)
 	}
 	return s
 }
 
-func (m Model) latencyStyle(ms int64, bg lipgloss.TerminalColor) lipgloss.Style {
-	var base lipgloss.TerminalColor
+func (m Model) latencyStyle(ms int64, bg lipgloss.Color) lipgloss.Style {
+	var hex string
 	var t float64
 	switch {
 	case ms < 200:
-		base = m.st.sparkSuccess
+		hex = m.st.sparkSuccess
 		t = float64(ms) / 200
 	case ms < 500:
-		base = m.st.sparkWarning
+		hex = m.st.sparkWarning
 		t = float64(ms-200) / 300
 	default:
-		base = m.st.sparkDanger
+		hex = m.st.sparkDanger
 		t = float64(ms-500) / 1500
 		if t > 1 {
 			t = 1
 		}
 	}
-	hex := trueColorHex(base)
 	s := lipgloss.NewStyle().Foreground(dimColor(hex, t))
 	return withBg(s, bg)
 }
 
-func (m Model) latencySparkline(latencies []time.Duration, statuses []bool, width int, bg lipgloss.TerminalColor) string {
+func (m Model) latencySparkline(latencies []time.Duration, statuses []bool, width int, bg lipgloss.Color) string {
 	if len(latencies) == 0 {
 		return withBg(m.st.subtleStyle, bg).Render(strings.Repeat("·", width))
 	}
@@ -115,7 +103,7 @@ func (m Model) latencySparkline(latencies []time.Duration, statuses []bool, widt
 	return sb.String()
 }
 
-func (m Model) heartbeatSparkline(statuses []bool, width int, bg lipgloss.TerminalColor) string {
+func (m Model) heartbeatSparkline(statuses []bool, width int, bg lipgloss.Color) string {
 	if len(statuses) == 0 {
 		return withBg(m.st.subtleStyle, bg).Render(strings.Repeat("·", width))
 	}
@@ -155,9 +143,10 @@ func resolveSparklineIndex(x, sparkWidth, dataLen int) int {
 	return offset + (x - padding)
 }
 
-func (m Model) groupSparkline(groupID int, width int, bg lipgloss.TerminalColor) string {
+func (m Model) groupSparkline(groupID int, width int, bg lipgloss.Color) string {
+	allSites := m.engine.GetAllSites()
 	var childStatuses [][]bool
-	for _, s := range m.sites {
+	for _, s := range allSites {
 		if s.ParentID == groupID && !s.Paused && !m.isMonitorInMaintenance(s.ID) {
 			hist, _ := m.engine.GetHistory(s.ID)
 			if len(hist.Statuses) > 0 {
@@ -208,8 +197,9 @@ func (m Model) groupSparkline(groupID int, width int, bg lipgloss.TerminalColor)
 }
 
 func (m Model) groupUptime(groupID int) string {
+	allSites := m.engine.GetAllSites()
 	var allStatuses [][]bool
-	for _, s := range m.sites {
+	for _, s := range allSites {
 		if s.ParentID == groupID && !s.Paused && !m.isMonitorInMaintenance(s.ID) {
 			hist, _ := m.engine.GetHistory(s.ID)
 			if len(hist.Statuses) > 0 {

internal/tui/sparkline_test.go

@@ -5,12 +5,10 @@ import (
 	"testing"
 	"time"
 	"unicode/utf8"
-
-	"github.com/charmbracelet/lipgloss"
 )
 
 func TestLatencySparkline_Empty(t *testing.T) {
-	got := styledModel.latencySparkline(nil, nil, 10, nil)
+	got := styledModel.latencySparkline(nil, nil, 10, "")
 	if !strings.Contains(got, "··········") {
 		t.Errorf("empty sparkline should be dots, got %q", got)
 	}
@@ -19,7 +17,7 @@ func TestLatencySparkline_Empty(t *testing.T) {
 func TestLatencySparkline_SingleValue(t *testing.T) {
 	latencies := []time.Duration{100 * time.Millisecond}
 	statuses := []bool{true}
-	got := styledModel.latencySparkline(latencies, statuses, 5, nil)
+	got := styledModel.latencySparkline(latencies, statuses, 5, "")
 	if len(got) == 0 {
 		t.Error("sparkline should not be empty")
 	}
@@ -35,7 +33,7 @@ func TestLatencySparkline_WidthTruncation(t *testing.T) {
 		latencies[i] = time.Duration(i*50) * time.Millisecond
 		statuses[i] = true
 	}
-	got := styledModel.latencySparkline(latencies, statuses, 5, nil)
+	got := styledModel.latencySparkline(latencies, statuses, 5, "")
 	if len(got) == 0 {
 		t.Error("sparkline should not be empty")
 	}
@@ -47,7 +45,7 @@ func TestLatencySparkline_WidthTruncation(t *testing.T) {
 func TestLatencySparkline_RelativeHeight(t *testing.T) {
 	latencies := []time.Duration{10 * time.Millisecond, 50 * time.Millisecond, 10 * time.Millisecond}
 	statuses := []bool{true, true, true}
-	out := stripANSI(styledModel.latencySparkline(latencies, statuses, 3, nil))
+	out := stripANSI(styledModel.latencySparkline(latencies, statuses, 3, ""))
 	runes := []rune(out)
 	if len(runes) < 3 {
 		t.Fatalf("expected 3 runes, got %d", len(runes))
@@ -59,14 +57,14 @@ func TestLatencySparkline_RelativeHeight(t *testing.T) {
 
 func TestLatencyStyle_BandsProduceDifferentColors(t *testing.T) {
 	st := newStyles(themeFlexokiDark)
-	st.sparkSuccess = lipgloss.Color("#00ff00")
-	st.sparkWarning = lipgloss.Color("#ffff00")
-	st.sparkDanger = lipgloss.Color("#ff0000")
+	st.sparkSuccess = "#00ff00"
+	st.sparkWarning = "#ffff00"
+	st.sparkDanger = "#ff0000"
 	m := Model{st: st}
 
-	green := m.latencyStyle(50, nil)
-	yellow := m.latencyStyle(300, nil)
-	red := m.latencyStyle(800, nil)
+	green := m.latencyStyle(50, "")
+	yellow := m.latencyStyle(300, "")
+	red := m.latencyStyle(800, "")
 
 	gfg := green.GetForeground()
 	yfg := yellow.GetForeground()
@@ -79,11 +77,11 @@ func TestLatencyStyle_BandsProduceDifferentColors(t *testing.T) {
 
 func TestLatencyStyle_BrightnessVariesWithinBand(t *testing.T) {
 	st := newStyles(themeFlexokiDark)
-	st.sparkSuccess = lipgloss.Color("#00ff00")
+	st.sparkSuccess = "#00ff00"
 	m := Model{st: st}
 
-	dim := m.latencyStyle(10, nil)
-	bright := m.latencyStyle(190, nil)
+	dim := m.latencyStyle(10, "")
+	bright := m.latencyStyle(190, "")
 
 	if dim.GetForeground() == bright.GetForeground() {
 		t.Error("10ms and 190ms should have different brightness within green band")
@@ -93,7 +91,7 @@ func TestLatencyStyle_BrightnessVariesWithinBand(t *testing.T) {
 func TestLatencySparkline_OutputWidth(t *testing.T) {
 	latencies := []time.Duration{100 * time.Millisecond, 200 * time.Millisecond, 300 * time.Millisecond}
 	statuses := []bool{true, true, true}
-	got := styledModel.latencySparkline(latencies, statuses, 5, nil)
+	got := styledModel.latencySparkline(latencies, statuses, 5, "")
 	count := utf8.RuneCountInString(stripANSI(got))
 	if count != 5 {
 		t.Errorf("expected 5 rune-width output, got %d from %q", count, got)
@@ -118,7 +116,7 @@ func stripANSI(s string) string {
 }
 
 func TestHeartbeatSparkline_Empty(t *testing.T) {
-	got := styledModel.heartbeatSparkline(nil, 10, nil)
+	got := styledModel.heartbeatSparkline(nil, 10, "")
 	if !strings.Contains(got, "··········") {
 		t.Errorf("empty heartbeat should be dots, got %q", got)
 	}
@@ -126,7 +124,7 @@ func TestHeartbeatSparkline_Empty(t *testing.T) {
 
 func TestHeartbeatSparkline_Mixed(t *testing.T) {
 	statuses := []bool{true, false, true, true, false}
-	got := styledModel.heartbeatSparkline(statuses, 5, nil)
+	got := styledModel.heartbeatSparkline(statuses, 5, "")
 	if len(got) == 0 {
 		t.Error("heartbeat sparkline should not be empty")
 	}
@@ -134,7 +132,7 @@ func TestHeartbeatSparkline_Mixed(t *testing.T) {
 
 func TestHeartbeatSparkline_PaddedWidth(t *testing.T) {
 	statuses := []bool{true, true}
-	got := styledModel.heartbeatSparkline(statuses, 5, nil)
+	got := styledModel.heartbeatSparkline(statuses, 5, "")
 	if !strings.Contains(got, "···") {
 		t.Errorf("should have dot padding for width > data, got %q", got)
 	}

internal/tui/tab_alerts.go

@@ -1,6 +1,7 @@
 package tui
 
 import (
+	"context"
 	"fmt"
 	neturl "net/url"
 	"sort"
@@ -184,7 +185,7 @@ func (m Model) viewAlertsTab() string {
 	nameW := widths[2]
 	cfgW := widths[4]
 
-	tbl := m.renderTable(
+	return m.renderTable(
 		headers,
 		len(m.alerts),
 		func(start, end int) [][]string {
@@ -205,101 +206,13 @@ func (m Model) viewAlertsTab() string {
 		},
 		widths, nil,
 	)
-
-	var totalSent, totalFail int
-	for _, a := range m.alerts {
-		h := m.engine.GetAlertHealth(a.ID)
-		totalSent += h.SendCount
-		totalFail += h.FailCount
-	}
-	types := make(map[string]bool)
-	for _, a := range m.alerts {
-		types[a.Type] = true
-	}
-	failLabel := "failures"
-	if totalFail == 1 {
-		failLabel = "failure"
-	}
-	summary := fmt.Sprintf("%d channels · %d types · %d sent · %d %s",
-		len(m.alerts), len(types), totalSent, totalFail, failLabel)
-
-	var detail string
-	if m.cursor < len(m.alerts) {
-		a := m.alerts[m.cursor]
-		h := m.engine.GetAlertHealth(a.ID)
-		detail = m.alertSelectedDetail(a, h)
-	}
-
-	return tbl + "\n  " + detail + "\n  " + m.st.subtleStyle.Render(summary)
-}
-
-func (m Model) alertSelectedDetail(a models.AlertConfig, h monitor.AlertHealth) string {
-	dot := m.st.subtleStyle.Render("  ·  ")
-	label := m.st.subtleStyle
-
-	parts := []string{fmtAlertType(a.Type)}
-	parts = append(parts, m.fmtAlertConfigFull(a))
-
-	if !h.LastSendAt.IsZero() {
-		if h.LastSendOK {
-			parts = append(parts, m.st.specialStyle.Render("●")+" "+label.Render("sent")+" "+m.fmtTimeAgo(h.LastSendAt))
-		} else {
-			parts = append(parts, m.st.dangerStyle.Render("●")+" "+label.Render("failed")+" "+m.fmtTimeAgo(h.LastSendAt))
-		}
-	}
-	if h.SendCount > 0 {
-		parts = append(parts, label.Render(fmt.Sprintf("%d sent, %d failed", h.SendCount, h.FailCount)))
-	}
-
-	return strings.Join(parts, dot)
-}
-
-func (m Model) fmtAlertConfigFull(alert models.AlertConfig) string {
-	switch alert.Type {
-	case "email":
-		host := alert.Settings["host"]
-		to := alert.Settings["to"]
-		if host != "" && to != "" {
-			return fmt.Sprintf("%s → %s", host, to)
-		}
-		if host != "" {
-			return host
-		}
-		return m.st.subtleStyle.Render("—")
-	case "ntfy":
-		topic := alert.Settings["topic"]
-		url := alert.Settings["url"]
-		if url != "" && topic != "" {
-			return fmt.Sprintf("%s/%s", url, topic)
-		}
-		return m.st.subtleStyle.Render("—")
-	case "telegram":
-		if id := alert.Settings["chat_id"]; id != "" {
-			return fmt.Sprintf("chat:%s", id)
-		}
-		return m.st.subtleStyle.Render("—")
-	case "gotify":
-		if url := alert.Settings["url"]; url != "" {
-			return url
-		}
-		return m.st.subtleStyle.Render("—")
-	default:
-		if val, ok := alert.Settings["url"]; ok && val != "" {
-			return maskWebhookURL(val)
-		}
-		return m.st.subtleStyle.Render("—")
-	}
 }
 
 func (m Model) viewAlertDetailPanel() string {
-	idx := m.cursor
-	if m.returnState == stateSettings {
-		idx = m.settingsCursor
-	}
-	if idx >= len(m.alerts) {
+	if m.cursor >= len(m.alerts) {
 		return ""
 	}
-	a := m.alerts[idx]
+	a := m.alerts[m.cursor]
 	h := m.engine.GetAlertHealth(a.ID)
 
 	var b strings.Builder
@@ -332,7 +245,7 @@ func (m Model) viewAlertDetailPanel() string {
 	}
 
 	b.WriteString(m.divider() + "\n")
-	b.WriteString(m.st.titleStyle.Render("  CONFIGURATION") + "\n")
+	b.WriteString(m.st.subtleStyle.Render("  CONFIGURATION") + "\n")
 	// Render through the same allowlist the backup export uses — this panel
 	// ends up in screen shares and asciinema recordings. Keys are sorted so
 	// rows don't reshuffle every render.
@@ -352,7 +265,7 @@ func (m Model) viewAlertDetailPanel() string {
 	}
 
 	b.WriteString(m.divider() + "\n")
-	b.WriteString(m.st.subtleStyle.Render("  [e] Edit  [t] Test  [q/Esc] Back"))
+	b.WriteString(m.st.subtleStyle.Render("  [q/Esc] Back  [e] Edit  [t] Test"))
 
 	return lipgloss.NewStyle().Padding(1, 2).Render(b.String())
 }
@@ -605,21 +518,15 @@ func (m *Model) submitAlertForm() tea.Cmd {
 	}
 
 	st := m.store
-	ctx := m.ctx
 	id := m.editID
 	name, aType := d.Name, d.AlertType
-	if m.returnState == stateSettings {
-		m.state = stateSettings
-	} else {
-		m.state = stateDashboard
-	}
-	m.returnState = 0
+	m.state = stateDashboard
 	if id > 0 {
 		return writeCmd("Update alert", func() error {
-			return st.UpdateAlert(ctx, id, name, aType, settings)
+			return st.UpdateAlert(context.Background(), id, name, aType, settings)
 		})
 	}
 	return writeCmd("Add alert", func() error {
-		return st.AddAlert(ctx, name, aType, settings)
+		return st.AddAlert(context.Background(), name, aType, settings)
 	})
 }

internal/tui/tab_logs.go

@@ -3,9 +3,6 @@ package tui
 import (
 	"fmt"
 	"strings"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-	"github.com/charmbracelet/lipgloss"
 )
 
 type logSeverity int
@@ -18,8 +15,8 @@ const (
 	severitySystem
 )
 
-func classifyLog(msg string) logSeverity {
-	lower := strings.ToLower(msg)
+func classifyLog(line string) logSeverity {
+	lower := strings.ToLower(line)
 	switch {
 	case strings.Contains(lower, "confirmed down"),
 		strings.Contains(lower, "is down"),
@@ -66,49 +63,67 @@ func (m Model) renderLogTag(sev logSeverity) string {
 	}
 }
 
-func (m Model) renderLogLine(entry models.LogEntry) string {
-	sev := classifyLog(entry.Message)
+func (m Model) renderLogLine(line string) string {
+	sev := classifyLog(line)
 	tag := m.renderLogTag(sev)
-	ts := m.st.subtleStyle.Render(entry.CreatedAt.Local().Format("01/02 15:04"))
-	return fmt.Sprintf("  %s  %s  %s", ts, tag, entry.Message)
-}
 
-func (m Model) viewLogsFullscreen() string {
-	header := "  " + m.st.titleStyle.Render("Logs") + "\n" + m.divider()
-
-	filterLabel := m.st.subtleStyle.Render("[f] All")
-	if m.logFilterImportant {
-		filterLabel = m.st.subtleStyle.Render("[f] Important only")
+	ts := ""
+	msg := line
+	if len(line) > 10 && line[0] == '[' {
+		if idx := strings.Index(line, "]"); idx > 0 && idx < 12 {
+			ts = m.st.subtleStyle.Render(line[1:idx])
+			msg = strings.TrimSpace(line[idx+1:])
+		}
 	}
-	countLabel := m.st.subtleStyle.Render(fmt.Sprintf("%d/%d", m.logShown, m.logTotal))
-	footer := m.divider() + "\n  " + filterLabel + "  " + countLabel + "  " + m.st.subtleStyle.Render("[q/Esc] Back")
 
-	m.logViewport.Width = m.termWidth - chromePadH
-	m.logViewport.Height = m.termHeight - 8
-
-	return lipgloss.NewStyle().Padding(1, 2).Render(
-		header + "\n" + m.logViewport.View() + "\n" + footer)
+	if ts != "" {
+		return fmt.Sprintf("  %s  %s  %s", ts, tag, msg)
+	}
+	return fmt.Sprintf("  %s  %s", tag, msg)
 }
 
+// refreshLogContent rebuilds the log viewport from the full engine log list,
+// filtering before windowing so the entry count and "(n hidden)" reflect all
+// logs, not just the visible viewport slice.
 func (m *Model) refreshLogContent() {
 	var rendered []string
 	total := 0
 	shown := 0
 
-	for _, entry := range m.engine.GetLogs() {
-		if strings.TrimSpace(entry.Message) == "" {
+	for _, line := range m.engine.GetLogs() {
+		if strings.TrimSpace(line) == "" {
 			continue
 		}
 		total++
-		sev := classifyLog(entry.Message)
+		sev := classifyLog(line)
 		if m.logFilterImportant && !isImportantLog(sev) {
 			continue
 		}
 		shown++
-		rendered = append(rendered, m.renderLogLine(entry))
+		rendered = append(rendered, m.renderLogLine(line))
 	}
 
 	m.logTotal = total
 	m.logShown = shown
 	m.logViewport.SetContent(strings.Join(rendered, "\n"))
 }
+
+func (m Model) viewLogsTab() string {
+	if m.logTotal == 0 {
+		return m.emptyState("No log entries yet.", "Logs appear as monitors run checks")
+	}
+
+	filterLabel := "All"
+	if m.logFilterImportant {
+		filterLabel = "Important"
+	}
+
+	header := m.st.subtleStyle.Render(fmt.Sprintf(
+		"  %d entries  Filter: %s", m.logShown, filterLabel))
+
+	if m.logFilterImportant && m.logShown < m.logTotal {
+		header += m.st.subtleStyle.Render(fmt.Sprintf("  (%d hidden)", m.logTotal-m.logShown))
+	}
+
+	return "\n" + header + "\n\n" + m.logViewport.View()
+}

internal/tui/tab_logs_sidebar.go

@@ -1,112 +0,0 @@
-package tui
-
-import (
-	"strings"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-	"github.com/charmbracelet/lipgloss"
-)
-
-func (m Model) renderCompactLogLine(entry models.LogEntry, maxW int) string {
-	sev := classifyLog(entry.Message)
-
-	var tag string
-	switch sev {
-	case severityDown:
-		tag = m.st.dangerStyle.Render("▼")
-	case severityUp:
-		tag = m.st.specialStyle.Render("▲")
-	case severityWarn:
-		tag = m.st.warnStyle.Render("◆")
-	case severitySystem:
-		tag = m.st.titleStyle.Render("●")
-	default:
-		tag = m.st.subtleStyle.Render("·")
-	}
-
-	ts := entry.CreatedAt.Local().Format("01/02 15:04")
-
-	msg := entry.Message
-	msg = strings.TrimPrefix(msg, "Monitor ")
-	msg = strings.TrimPrefix(msg, "Push ")
-
-	prefixW := len(ts) + 4
-	msgW := maxW - prefixW
-	if msgW < 5 {
-		msgW = 5
-	}
-	msg = limitStr(msg, msgW)
-
-	return " " + m.st.subtleStyle.Render(ts) + " " + tag + " " + msg
-}
-
-func (m Model) viewLogsStrip(width, maxLines int) string {
-	logs := m.engine.GetLogs()
-	if len(logs) == 0 {
-		return m.st.subtleStyle.Render("  No logs yet")
-	}
-
-	style := lipgloss.NewStyle().Width(width).MaxWidth(width)
-
-	var all []string
-	for _, entry := range logs {
-		if strings.TrimSpace(entry.Message) == "" {
-			continue
-		}
-		if m.logFilterImportant && !isImportantLog(classifyLog(entry.Message)) {
-			continue
-		}
-		all = append(all, m.renderCompactLogLine(entry, width))
-	}
-
-	start := m.logScrollOffset
-	if start > len(all) {
-		start = len(all)
-	}
-	end := start + maxLines
-	if end > len(all) {
-		end = len(all)
-	}
-	visible := all[start:end]
-
-	return style.Render(strings.Join(visible, "\n"))
-}
-
-func (m Model) filteredLogCount() int {
-	count := 0
-	for _, entry := range m.engine.GetLogs() {
-		if strings.TrimSpace(entry.Message) == "" {
-			continue
-		}
-		if m.logFilterImportant && !isImportantLog(classifyLog(entry.Message)) {
-			continue
-		}
-		count++
-	}
-	return count
-}
-
-func (m *Model) scrollLogs(delta int) {
-	logs := m.engine.GetLogs()
-	total := 0
-	for _, entry := range logs {
-		if strings.TrimSpace(entry.Message) == "" {
-			continue
-		}
-		if m.logFilterImportant && !isImportantLog(classifyLog(entry.Message)) {
-			continue
-		}
-		total++
-	}
-
-	m.logScrollOffset += delta
-	if m.logScrollOffset < 0 {
-		m.logScrollOffset = 0
-	}
-	if m.logScrollOffset > total-1 {
-		m.logScrollOffset = total - 1
-	}
-	if m.logScrollOffset < 0 {
-		m.logScrollOffset = 0
-	}
-}

internal/tui/tab_maint.go

@@ -1,6 +1,7 @@
 package tui
 
 import (
+	"context"
 	"fmt"
 	"strconv"
 	"time"
@@ -20,37 +21,115 @@ type maintFormData struct {
 	CustomHours string
 }
 
-func (m Model) isMonitorInMaintenance(monitorID int) bool {
-	return m.maintSet[monitorID]
+func (m Model) fmtMaintStatus(mw models.MaintenanceWindow) string {
+	now := time.Now()
+	if mw.StartTime.After(now) {
+		return m.st.warnStyle.Render("SCHEDULED")
+	}
+	if !mw.EndTime.IsZero() && mw.EndTime.Before(now) {
+		return m.st.subtleStyle.Render("ENDED")
+	}
+	return m.st.specialStyle.Render("ACTIVE")
 }
 
-func (m *Model) buildMaintSet() {
-	set := make(map[int]bool)
+func (m Model) fmtMaintType(t string) string {
+	if t == "incident" {
+		return m.st.dangerStyle.Render("incident")
+	}
+	return m.st.maintStyle.Render("maintenance")
+}
+
+func fmtMaintMonitorW(monitorID int, sites []models.Site, maxW int) string {
+	if monitorID == 0 {
+		return "All"
+	}
+	for _, s := range sites {
+		if s.ID == monitorID {
+			return limitStr(s.Name, maxW)
+		}
+	}
+	return fmt.Sprintf("#%d", monitorID)
+}
+
+func (m Model) fmtMaintTime(t time.Time, colW int) string {
+	if t.IsZero() {
+		return m.st.subtleStyle.Render("—")
+	}
 	now := time.Now()
+	if t.Year() == now.Year() && t.YearDay() == now.YearDay() {
+		return t.Format("15:04")
+	}
+	if colW >= 14 {
+		return t.Format("15:04 Jan 02")
+	}
+	return t.Format("Jan 02")
+}
+
+func (m Model) isMonitorInMaintenance(monitorID int) bool {
 	for _, mw := range m.maintenanceWindows {
 		if mw.Type != "maintenance" {
 			continue
 		}
+		now := time.Now()
 		if mw.StartTime.After(now) {
 			continue
 		}
 		if !mw.EndTime.IsZero() && mw.EndTime.Before(now) {
 			continue
 		}
-		if mw.MonitorID == 0 {
-			for _, s := range m.sites {
-				set[s.ID] = true
-			}
-			break
+		if mw.MonitorID == 0 || mw.MonitorID == monitorID {
+			return true
 		}
-		set[mw.MonitorID] = true
 		for _, s := range m.sites {
-			if s.ParentID == mw.MonitorID {
-				set[s.ID] = true
+			if s.ID == monitorID && s.ParentID > 0 && mw.MonitorID == s.ParentID {
+				return true
 			}
 		}
 	}
-	m.maintSet = set
+	return false
+}
+
+func (m Model) viewMaintTab() string {
+	if len(m.maintenanceWindows) == 0 {
+		return m.emptyState("No maintenance windows or incidents.", "[n] Create one")
+	}
+
+	var headers []string
+	var widths []int
+	if m.isWide() {
+		headers = []string{"#", "TITLE", "TYPE", "MONITORS", "STATUS", "STARTED", "ENDS"}
+		widths = []int{4, 24, 14, 22, 12, 16, 16}
+	} else {
+		headers = []string{"#", "TITLE", "TYPE", "MON", "ST", "START", "ENDS"}
+		widths = []int{4, 14, 13, 14, 11, 14, 14}
+	}
+	titleW := widths[1]
+	monW := widths[3]
+	timeW := widths[5]
+
+	return m.renderTable(
+		headers,
+		len(m.maintenanceWindows),
+		func(start, end int) [][]string {
+			var rows [][]string
+			allSites := m.engine.GetAllSites()
+			for i := start; i < end; i++ {
+				mw := m.maintenanceWindows[i]
+				rows = append(rows, []string{
+					strconv.Itoa(i + 1),
+					m.zones.Mark(fmt.Sprintf("maint-%d", i), limitStr(mw.Title, titleW-2)),
+					m.fmtMaintType(mw.Type),
+					fmtMaintMonitorW(mw.MonitorID, allSites, monW-2),
+					m.fmtMaintStatus(mw),
+					m.fmtMaintTime(mw.StartTime, timeW),
+					m.fmtMaintTime(mw.EndTime, timeW),
+				})
+			}
+			return rows
+		},
+		widths,
+		nil,
+	)
 }
 
 func (m *Model) initMaintHuhForm() tea.Cmd {
@@ -160,13 +239,8 @@ func (m *Model) submitMaintForm() tea.Cmd {
 	}
 
 	st := m.store
-	ctx := m.ctx
 	m.state = stateDashboard
 	return writeCmd("Add maintenance window", func() error {
-		overlaps, _ := st.GetOverlappingMaintenanceWindows(ctx, mw.MonitorID, mw.StartTime, mw.EndTime)
-		if len(overlaps) > 0 {
-			_ = st.SaveLog(ctx, fmt.Sprintf("Overlap: new window %q overlaps with existing %q", mw.Title, overlaps[0].Title))
-		}
-		return st.AddMaintenanceWindow(ctx, mw)
+		return st.AddMaintenanceWindow(context.Background(), mw)
 	})
 }

internal/tui/tab_nodes.go

@@ -1,28 +1,26 @@
 package tui
 
 import (
-	"fmt"
-	"strings"
 	"time"
 )
 
 func (m Model) viewNodesTab() string {
 	if len(m.nodes) == 0 {
-		return m.emptyState("No probe nodes connected.", "Probes auto-register on connection.")
+		return m.emptyState("No probe nodes connected.", "")
 	}
 
 	var headers []string
 	var widths []int
 	if m.isWide() {
-		headers = []string{"#", "NAME", "REGION", "LAST SEEN", "VERSION", "STATUS"}
-		widths = []int{4, 24, 14, 16, 12, 10}
+		headers = []string{"NAME", "REGION", "LAST SEEN", "VERSION", "STATUS"}
+		widths = []int{24, 14, 16, 12, 10}
 	} else {
-		headers = []string{"#", "NAME", "REGION", "SEEN", "VER", "STATUS"}
-		widths = []int{4, 16, 10, 10, 8, 8}
+		headers = []string{"NAME", "REGION", "SEEN", "VER", "STATUS"}
+		widths = []int{16, 10, 10, 8, 8}
 	}
-	nameW := widths[1]
+	nameW := widths[0]
 
-	tbl := m.renderTable(
+	return m.renderTable(
 		headers,
 		len(m.nodes),
 		func(start, end int) [][]string {
@@ -43,37 +41,13 @@ func (m Model) viewNodesTab() string {
 					version = m.st.subtleStyle.Render("—")
 				}
 				status := m.fmtNodeStatus(node.LastSeen)
-				rows = append(rows, []string{fmt.Sprintf("%d", i+1), name, region, lastSeen, version, status})
+				rows = append(rows, []string{name, region, lastSeen, version, status})
 			}
 			return rows
 		},
 		widths,
 		nil,
 	)
-
-	var online int
-	regions := make(map[string]bool)
-	var leader string
-	for _, n := range m.nodes {
-		if time.Since(n.LastSeen) < nodeOnlineThreshold {
-			online++
-		}
-		if n.Region != "" {
-			regions[n.Region] = true
-		}
-		if n.Name == "leader" {
-			leader = n.ID
-		}
-	}
-	parts := []string{fmt.Sprintf("%d/%d online", online, len(m.nodes))}
-	if leader != "" {
-		parts = append(parts, fmt.Sprintf("leader: %s", leader))
-	}
-	if len(regions) > 0 {
-		parts = append(parts, fmt.Sprintf("%d regions", len(regions)))
-	}
-
-	return tbl + "\n  " + m.st.subtleStyle.Render(strings.Join(parts, " · "))
 }
 
 func (m Model) fmtNodeStatus(lastSeen time.Time) string {
@@ -81,10 +55,10 @@ func (m Model) fmtNodeStatus(lastSeen time.Time) string {
 		return m.st.subtleStyle.Render("UNKNOWN")
 	}
 	ago := time.Since(lastSeen)
-	if ago < nodeOnlineThreshold {
+	if ago < 60*time.Second {
 		return m.st.specialStyle.Render("ONLINE")
 	}
-	if ago < nodeStaleThreshold {
+	if ago < 5*time.Minute {
 		return m.st.warnStyle.Render("STALE")
 	}
 	return m.st.dangerStyle.Render("OFFLINE")

internal/tui/tab_settings.go

@@ -1,56 +0,0 @@
-package tui
-
-import (
-	"fmt"
-
-	"github.com/charmbracelet/lipgloss"
-)
-
-func (m Model) viewSettingsTab() string {
-	sections := []string{"Alerts", "Nodes"}
-	if m.isAdmin {
-		sections = append(sections, "Users")
-	}
-
-	var tabs []string
-	for i, name := range sections {
-		var rendered string
-		if i == m.settingsSection {
-			rendered = m.st.activeTab.Render(name)
-		} else {
-			rendered = m.st.inactiveTab.Render(name)
-		}
-		tabs = append(tabs, m.zones.Mark(fmt.Sprintf("section-%d", i), rendered))
-	}
-	header := lipgloss.JoinHorizontal(lipgloss.Top, tabs...)
-
-	var content string
-	switch m.settingsSection {
-	case sectionAlerts:
-		content = m.viewAlertsTab()
-	case sectionNodes:
-		content = m.viewNodesTab()
-	case sectionUsers:
-		if m.isAdmin {
-			content = m.viewUsersTab()
-		}
-	}
-
-	return header + "\n" + content
-}
-
-func (m *Model) switchSettingsSection(idx int) {
-	max := 1
-	if m.isAdmin {
-		max = 2
-	}
-	if idx > max {
-		idx = 0
-	}
-	if idx < 0 {
-		idx = max
-	}
-	m.settingsSection = idx
-	m.cursor = 0
-	m.tableOffset = 0
-}

internal/tui/tab_sites.go

@@ -1,6 +1,7 @@
 package tui
 
 import (
+	"context"
 	"fmt"
 	"net/url"
 	"strconv"
@@ -34,9 +35,10 @@ type siteFormData struct {
 type colKey int
 
 const (
-	colDot colKey = iota
+	colNum colKey = iota
 	colName
 	colType
+	colStatus
 	colLatency
 	colUptime
 	colHistory
@@ -54,18 +56,17 @@ type columnDef struct {
 }
 
 var siteColumns = []columnDef{
-	{colDot, "", "", 3, 3, 0},
+	{colNum, "#", "#", 4, 4, 0},
 	{colName, "NAME", "NAME", 0, 0, 0},
-	{colType, "TYPE", "TYPE", 10, 8, 0},
+	{colType, "TYPE", "TYPE", 10, 8, mediumBreakpoint},
+	{colStatus, "STATUS", "STATUS", 10, 10, 0},
 	{colLatency, "LATENCY", "LAT", 10, 7, 0},
-	{colUptime, "UPTIME", "UP%", 8, 8, 0},
-	{colHistory, "HISTORY", "HISTORY", 0, 0, 0},
-	{colSSL, "SSL", "SSL", 7, 5, 0},
-	{colRetries, "RETRIES", "RT", 9, 5, 0},
+	{colUptime, "UPTIME", "UP%", 8, 8, mediumBreakpoint},
+	{colHistory, "HISTORY", "HISTORY", 0, 0, mediumBreakpoint},
+	{colSSL, "SSL", "SSL", 7, 5, wideBreakpoint},
+	{colRetries, "RETRIES", "RT", 9, 5, wideBreakpoint},
 }
 
-var columnDropOrder = []colKey{colHistory, colRetries, colSSL, colUptime, colType}
-
 type tableLayout struct {
 	nameW, sparkW int
 	headers       []string
@@ -76,51 +77,13 @@ type tableLayout struct {
 func (m Model) computeLayout() tableLayout {
 	wide := m.isWide()
 
-	cw := m.contentWidth
-	if cw == 0 {
-		cw = m.termWidth
-	}
-
-	dropped := make(map[colKey]bool)
-	minNameW := 20
-	minSparkW := 12
-
-	for attempt := 0; attempt <= len(columnDropOrder); attempt++ {
-		var fixed int
-		var flexCount int
-		for _, c := range siteColumns {
-			if dropped[c.key] {
-				continue
-			}
-			w := c.narrowW
-			if wide {
-				w = c.wideW
-			}
-			if w > 0 {
-				fixed += w
-			} else {
-				flexCount++
-			}
-		}
-		numCols := len(siteColumns) - len(dropped)
-		borderOverhead := 2 + (numCols - 1)
-		avail := cw - chromePadH - 2 - borderOverhead - fixed
-		minFlex := minNameW
-		if flexCount > 1 {
-			minFlex += minSparkW
-		}
-		if avail >= minFlex || attempt >= len(columnDropOrder) {
-			break
-		}
-		dropped[columnDropOrder[attempt]] = true
-	}
-
 	var active []colKey
 	var headers []string
 	var widths []int
 	var fixed int
+
 	for _, c := range siteColumns {
-		if dropped[c.key] {
+		if c.minTerm > 0 && m.termWidth < c.minTerm {
 			continue
 		}
 		active = append(active, c.key)
@@ -139,37 +102,9 @@ func (m Model) computeLayout() tableLayout {
 		}
 	}
 
-	sortColMap := map[int]colKey{
-		sortStatus:  colDot,
-		sortName:    colName,
-		sortLatency: colLatency,
-	}
-	sortableKeys := map[colKey]string{
-		colDot:     "sort-status",
-		colName:    "sort-name",
-		colLatency: "sort-latency",
-	}
-	if sortedKey, ok := sortColMap[m.sortColumn]; ok {
-		arrow := "▼"
-		if m.sortAsc {
-			arrow = "▲"
-		}
-		for i, k := range active {
-			if k == sortedKey {
-				headers[i] = headers[i] + arrow
-				break
-			}
-		}
-	}
-	for i, k := range active {
-		if zoneID, ok := sortableKeys[k]; ok {
-			headers[i] = m.zones.Mark(zoneID, headers[i])
-		}
-	}
-
 	numCols := len(headers)
 	borderOverhead := 2 + (numCols - 1)
-	avail := cw - chromePadH - 2 - borderOverhead - fixed
+	avail := m.termWidth - chromePadH - 2 - borderOverhead - fixed
 	if avail < 20 {
 		avail = 20
 	}
@@ -269,7 +204,7 @@ func (m Model) viewSitesTab() string {
 			for i := start; i < end; i++ {
 				site := m.sites[i]
 				rowIdx := i - start
-				var rowBg lipgloss.TerminalColor
+				var rowBg lipgloss.Color
 				if i == m.cursor {
 					rowBg = m.theme.SelectedBg
 				} else if rowIdx%2 == 1 {
@@ -279,11 +214,11 @@ func (m Model) viewSitesTab() string {
 				if site.Type == "group" {
 					groupRows[i-start] = true
 					icon := typeIcon("group", m.collapsed[site.ID])
-					inMaint := m.isMonitorInMaintenance(site.ID)
 					cells := map[colKey]string{
-						colDot:     m.fmtStatusDot(site.Status, site.Paused, inMaint),
+						colNum:     strconv.Itoa(i + 1),
 						colName:    m.zones.Mark(fmt.Sprintf("site-%d", i), icon+" "+limitStr(site.Name, nameW-4)),
 						colType:    "group",
+						colStatus:  m.fmtStatus(site.Status, site.Paused, m.isMonitorInMaintenance(site.ID)),
 						colLatency: m.st.subtleStyle.Render("—"),
 						colUptime:  m.groupUptime(site.ID),
 						colHistory: m.groupSparkline(site.ID, sparkWidth, rowBg),
@@ -327,13 +262,13 @@ func (m Model) viewSitesTab() string {
 					spark = m.latencySparkline(hist.Latencies, hist.Statuses, sparkWidth, rowBg)
 				}
 
-				inMaint := m.isMonitorInMaintenance(site.ID)
 				cells := map[colKey]string{
-					colDot:     m.fmtStatusDot(site.Status, site.Paused, inMaint),
+					colNum:     strconv.Itoa(i + 1),
 					colName:    m.zones.Mark(fmt.Sprintf("site-%d", i), name),
 					colType:    typeIcon(site.Type, false) + " " + site.Type,
+					colStatus:  m.fmtStatus(site.Status, site.Paused, m.isMonitorInMaintenance(site.ID)),
 					colLatency: m.fmtLatency(site.Latency),
-					colUptime:  m.fmtUptimeMaint(hist.Statuses, site.ID),
+					colUptime:  m.fmtUptime(hist.Statuses),
 					colHistory: spark,
 					colSSL:     m.fmtSSL(site),
 					colRetries: m.fmtRetries(site),
@@ -391,104 +326,101 @@ func (m *Model) initSiteHuhForm() tea.Cmd {
 		}
 	}
 
-	return m.rebuildSiteForm()
-}
-
-func (m *Model) rebuildSiteForm() tea.Cmd {
-	groups := m.buildSiteFormGroups()
-	m.huhForm = huh.NewForm(groups...).WithTheme(m.theme.HuhTheme())
-	if m.termWidth > 0 {
-		m.huhForm.WithWidth(m.termWidth)
-	}
-	formHeight := m.termHeight - 7
-	if formHeight < 5 {
-		formHeight = 5
-	}
-	m.huhForm.WithHeight(formHeight)
-	m.lastSiteType = m.siteFormData.SiteType
-	return m.huhForm.Init()
-}
-
-func (m *Model) siteFormOptions() (alertOpts, groupOpts []huh.Option[string]) {
-	alertOpts = []huh.Option[string]{huh.NewOption("None", "0")}
+	// m.alerts is the tab-data cache (≤5s stale) — no store IO in Update.
+	alertOpts := []huh.Option[string]{huh.NewOption("None", "0")}
 	for _, a := range m.alerts {
 		alertOpts = append(alertOpts, huh.NewOption(
 			fmt.Sprintf("%s (%s)", a.Name, a.Type),
 			strconv.Itoa(a.ID),
 		))
 	}
-	groupOpts = []huh.Option[string]{huh.NewOption("None", "0")}
+
+	groupOpts := []huh.Option[string]{huh.NewOption("None", "0")}
 	for _, s := range m.sites {
 		if s.Type == "group" && s.ID != m.editID {
 			groupOpts = append(groupOpts, huh.NewOption(s.Name, strconv.Itoa(s.ID)))
 		}
 	}
-	return
-}
 
-func (m *Model) buildSiteFormGroups() []*huh.Group {
-	d := m.siteFormData
-	alertOpts, groupOpts := m.siteFormOptions()
-
-	// Page 1 — Monitor Setup: core fields + type-specific target
-	setup := []huh.Field{
-		huh.NewInput().Title("Monitor Name").
-			Placeholder("My Service").
-			Value(&d.Name).
-			Validate(func(s string) error {
-				if s == "" {
-					return fmt.Errorf("name is required")
-				}
-				return nil
-			}),
-		huh.NewSelect[string]().Title("Monitor Type").
-			Options(
-				huh.NewOption("HTTP/HTTPS", "http"),
-				huh.NewOption("Push / Heartbeat", "push"),
-				huh.NewOption("Ping (ICMP)", "ping"),
-				huh.NewOption("TCP Port", "port"),
-				huh.NewOption("DNS", "dns"),
-				huh.NewOption("Group", "group"),
-			).Value(&d.SiteType),
-		huh.NewSelect[string]().Title("Alert Channel").
-			Options(alertOpts...).
-			Value(&d.AlertID),
-	}
-
-	switch d.SiteType {
-	case "http":
-		setup = append(setup, huh.NewInput().Title("URL").
-			Placeholder("https://example.com").
-			Value(&d.URL).
-			Validate(func(s string) error {
-				if s == "" {
-					return fmt.Errorf("URL is required")
-				}
-				u, err := url.Parse(s)
-				if err != nil {
-					return fmt.Errorf("invalid URL")
-				}
-				if u.Scheme != "http" && u.Scheme != "https" {
-					return fmt.Errorf("URL must start with http:// or https://")
-				}
-				if u.Host == "" {
-					return fmt.Errorf("URL must include a host")
-				}
-				return nil
-			}))
-	case "ping", "dns":
-		setup = append(setup, huh.NewInput().Title("Hostname / IP").
-			Placeholder("10.0.0.1").
-			Value(&d.Hostname))
-	case "port":
-		setup = append(setup,
+	m.huhForm = huh.NewForm(
+		huh.NewGroup(
+			huh.NewInput().Title("Monitor Name").
+				Placeholder("My Service").
+				Value(&m.siteFormData.Name).
+				Validate(func(s string) error {
+					if s == "" {
+						return fmt.Errorf("name is required")
+					}
+					return nil
+				}),
+			huh.NewSelect[string]().Title("Monitor Type").
+				Options(
+					huh.NewOption("HTTP/HTTPS", "http"),
+					huh.NewOption("Push / Heartbeat", "push"),
+					huh.NewOption("Ping (ICMP)", "ping"),
+					huh.NewOption("TCP Port", "port"),
+					huh.NewOption("DNS", "dns"),
+					huh.NewOption("Group", "group"),
+				).Value(&m.siteFormData.SiteType),
+			huh.NewSelect[string]().Title("Alert Channel").
+				Options(alertOpts...).
+				Value(&m.siteFormData.AlertID),
+		).Title("Monitor Settings"),
+		huh.NewGroup(
+			huh.NewInput().Title("URL").
+				Placeholder("https://example.com").
+				Description("Required for HTTP monitors").
+				Value(&m.siteFormData.URL).
+				Validate(func(s string) error {
+					if m.siteFormData.SiteType != "http" {
+						return nil
+					}
+					if s == "" {
+						return fmt.Errorf("URL is required for HTTP monitors")
+					}
+					u, err := url.Parse(s)
+					if err != nil {
+						return fmt.Errorf("invalid URL")
+					}
+					if u.Scheme != "http" && u.Scheme != "https" {
+						return fmt.Errorf("URL must start with http:// or https://")
+					}
+					if u.Host == "" {
+						return fmt.Errorf("URL must include a host")
+					}
+					return nil
+				}),
+			huh.NewInput().Title("Check Interval (seconds)").
+				Placeholder("60").
+				Value(&m.siteFormData.Interval).
+				Validate(func(s string) error {
+					if m.siteFormData.SiteType == "group" {
+						return nil
+					}
+					v, err := strconv.Atoi(s)
+					if err != nil {
+						return fmt.Errorf("must be a number")
+					}
+					if v < 5 {
+						return fmt.Errorf("minimum interval is 5 seconds")
+					}
+					return nil
+				}),
+			huh.NewSelect[string]().Title("Parent Group").
+				Options(groupOpts...).
+				Value(&m.siteFormData.GroupID),
 			huh.NewInput().Title("Hostname / IP").
 				Placeholder("10.0.0.1").
-				Value(&d.Hostname),
+				Description("Target for ping/port/DNS monitors").
+				Value(&m.siteFormData.Hostname),
 			huh.NewInput().Title("Port").
-				Placeholder("443").
-				Value(&d.Port).
+				Placeholder("0").
+				Description("Target port for TCP port monitors").
+				Value(&m.siteFormData.Port).
 				Validate(func(s string) error {
+					if m.siteFormData.SiteType != "port" {
+						return nil
+					}
 					v, err := strconv.Atoi(s)
 					if err != nil {
 						return fmt.Errorf("must be a number")
@@ -497,20 +429,34 @@ func (m *Model) buildSiteFormGroups() []*huh.Group {
 						return fmt.Errorf("port must be 1-65535")
 					}
 					return nil
-				}))
-	}
-
-	groups := []*huh.Group{huh.NewGroup(setup...).Title("Monitor Setup")}
-
-	if d.SiteType == "group" {
-		return groups
-	}
-
-	// Page 2 — Configuration: type-specific options + shared defaults
-	var config []huh.Field
-
-	if d.SiteType == "http" {
-		config = append(config,
+				}),
+			huh.NewInput().Title("Timeout (seconds)").
+				Placeholder("5").
+				Value(&m.siteFormData.Timeout).
+				Validate(func(s string) error {
+					if m.siteFormData.SiteType == "group" {
+						return nil
+					}
+					v, err := strconv.Atoi(s)
+					if err != nil {
+						return fmt.Errorf("must be a number")
+					}
+					if v < 1 || v > 300 {
+						return fmt.Errorf("timeout must be 1-300 seconds")
+					}
+					return nil
+				}),
+			huh.NewInput().Title("Description").
+				Placeholder("Optional description").
+				Value(&m.siteFormData.Description),
+			huh.NewInput().Title("Probe Regions").
+				Placeholder("us-east, eu-west (empty = all)").
+				Description("Comma-separated regions for distributed probing").
+				Value(&m.siteFormData.Regions),
+		).Title("Connection").WithHideFunc(func() bool {
+			return m.siteFormData.SiteType == "group"
+		}),
+		huh.NewGroup(
 			huh.NewSelect[string]().Title("HTTP Method").
 				Options(
 					huh.NewOption("GET", "GET"),
@@ -520,75 +466,22 @@ func (m *Model) buildSiteFormGroups() []*huh.Group {
 					huh.NewOption("DELETE", "DELETE"),
 					huh.NewOption("HEAD", "HEAD"),
 					huh.NewOption("OPTIONS", "OPTIONS"),
-				).Value(&d.Method),
+				).Value(&m.siteFormData.Method),
 			huh.NewInput().Title("Accepted Status Codes").
 				Placeholder("200-299").
 				Description("Ranges (200-299) and singles (301) separated by commas").
-				Value(&d.AcceptedCodes),
-		)
-	}
-
-	config = append(config,
-		huh.NewInput().Title("Check Interval (seconds)").
-			Placeholder("60").
-			Value(&d.Interval).
-			Validate(func(s string) error {
-				v, err := strconv.Atoi(s)
-				if err != nil {
-					return fmt.Errorf("must be a number")
-				}
-				if v < 5 {
-					return fmt.Errorf("minimum interval is 5 seconds")
-				}
-				return nil
-			}),
-		huh.NewInput().Title("Timeout (seconds)").
-			Placeholder("5").
-			Value(&d.Timeout).
-			Validate(func(s string) error {
-				v, err := strconv.Atoi(s)
-				if err != nil {
-					return fmt.Errorf("must be a number")
-				}
-				if v < 1 || v > 300 {
-					return fmt.Errorf("timeout must be 1-300 seconds")
-				}
-				return nil
-			}),
-		huh.NewInput().Title("Max Retries Before Alert").
-			Placeholder("0").
-			Value(&d.Retries).
-			Validate(func(s string) error {
-				v, err := strconv.Atoi(s)
-				if err != nil {
-					return fmt.Errorf("must be a number")
-				}
-				if v < 0 {
-					return fmt.Errorf("retries cannot be negative")
-				}
-				return nil
-			}),
-		huh.NewSelect[string]().Title("Parent Group").
-			Options(groupOpts...).
-			Value(&d.GroupID),
-		huh.NewInput().Title("Description").
-			Placeholder("Optional description").
-			Value(&d.Description),
-		huh.NewInput().Title("Probe Regions").
-			Placeholder("us-east, eu-west (empty = all)").
-			Description("Comma-separated regions for distributed probing").
-			Value(&d.Regions),
-	)
-
-	if d.SiteType == "http" {
-		config = append(config,
+				Value(&m.siteFormData.AcceptedCodes),
+		).Title("HTTP Settings").WithHideFunc(func() bool {
+			return m.siteFormData.SiteType != "http"
+		}),
+		huh.NewGroup(
 			huh.NewConfirm().Title("Monitor SSL Certificate?").
-				Value(&d.CheckSSL),
+				Value(&m.siteFormData.CheckSSL),
 			huh.NewInput().Title("SSL Warning Threshold (days)").
 				Placeholder("7").
-				Value(&d.Threshold).
+				Value(&m.siteFormData.Threshold).
 				Validate(func(s string) error {
-					if !d.CheckSSL {
+					if !m.siteFormData.CheckSSL {
 						return nil
 					}
 					v, err := strconv.Atoi(s)
@@ -600,13 +493,30 @@ func (m *Model) buildSiteFormGroups() []*huh.Group {
 					}
 					return nil
 				}),
+			huh.NewInput().Title("Max Retries Before Alert").
+				Placeholder("0").
+				Value(&m.siteFormData.Retries).
+				Validate(func(s string) error {
+					if m.siteFormData.SiteType == "group" {
+						return nil
+					}
+					v, err := strconv.Atoi(s)
+					if err != nil {
+						return fmt.Errorf("must be a number")
+					}
+					if v < 0 {
+						return fmt.Errorf("retries cannot be negative")
+					}
+					return nil
+				}),
 			huh.NewConfirm().Title("Ignore TLS Errors?").
-				Value(&d.IgnoreTLS),
-		)
-	}
+				Value(&m.siteFormData.IgnoreTLS),
+		).Title("Advanced").WithHideFunc(func() bool {
+			return m.siteFormData.SiteType == "group"
+		}),
+	).WithTheme(m.theme.HuhTheme())
 
-	groups = append(groups, huh.NewGroup(config...).Title("Configuration"))
-	return groups
+	return m.huhForm.Init()
 }
 
 func (m *Model) submitSiteForm() tea.Cmd {
@@ -647,11 +557,10 @@ func (m *Model) submitSiteForm() tea.Cmd {
 	}
 
 	st := m.store
-	ctx := m.ctx
 	m.state = stateDashboard
 	if m.editID > 0 {
 		m.engine.UpdateSiteConfig(cfg)
-		return writeCmd("Update site", func() error { return st.UpdateSite(ctx, cfg) })
+		return writeCmd("Update site", func() error { return st.UpdateSite(context.Background(), cfg) })
 	}
-	return writeCmd("Add site", func() error { return st.AddSite(ctx, cfg) })
+	return writeCmd("Add site", func() error { return st.AddSite(context.Background(), cfg) })
 }

internal/tui/tab_users.go

@@ -1,6 +1,7 @@
 package tui
 
 import (
+	"context"
 	"fmt"
 
 	tea "github.com/charmbracelet/bubbletea"
@@ -113,21 +114,15 @@ func (m *Model) initUserHuhForm() tea.Cmd {
 func (m *Model) submitUserForm() tea.Cmd {
 	d := m.userFormData
 	st := m.store
-	ctx := m.ctx
 	id := m.editID
 	username, key, role := d.Username, d.PublicKey, d.Role
-	if m.returnState == stateSettings {
-		m.state = stateSettings
-	} else {
-		m.state = stateDashboard
-	}
-	m.returnState = 0
+	m.state = stateUsers
 	if id > 0 {
 		return writeCmd("Update user", func() error {
-			return st.UpdateUser(ctx, id, username, key, role)
+			return st.UpdateUser(context.Background(), id, username, key, role)
 		})
 	}
 	return writeCmd("Add user", func() error {
-		return st.AddUser(ctx, username, key, role)
+		return st.AddUser(context.Background(), username, key, role)
 	})
 }

internal/tui/table_helpers.go

@@ -13,11 +13,7 @@ const (
 )
 
 func (m Model) isWide() bool {
-	w := m.contentWidth
-	if w == 0 {
-		w = m.termWidth
-	}
-	return w >= wideBreakpoint
+	return m.termWidth >= wideBreakpoint
 }
 
 func (m Model) renderTable(headers []string, items int, buildRows func(start, end int) [][]string, colWidths []int, styleOverride StyleOverride) string {
@@ -39,11 +35,7 @@ func (m Model) renderTable(headers []string, items int, buildRows func(start, en
 	}
 	borderOverhead := 2 + len(colWidths) - 1
 	tableWidth := colTotal + borderOverhead
-	cw := m.contentWidth
-	if cw == 0 {
-		cw = m.termWidth
-	}
-	maxWidth := cw - chromePadH - 2
+	maxWidth := m.termWidth - chromePadH - 2
 	if tableWidth > maxWidth {
 		tableWidth = maxWidth
 	}
@@ -52,7 +44,8 @@ func (m Model) renderTable(headers []string, items int, buildRows func(start, en
 	}
 
 	t := table.New().
-		Border(lipgloss.HiddenBorder()).
+		Border(lipgloss.RoundedBorder()).
+		BorderStyle(m.st.tableBorderStyle).
 		Width(tableWidth).
 		Headers(headers...).
 		Rows(rows...).
@@ -93,5 +86,5 @@ func (m Model) renderTable(headers []string, items int, buildRows func(start, en
 			return base
 		})
 
-	return t.Render()
+	return "\n" + t.Render()
 }

internal/tui/theme.go

@@ -5,200 +5,148 @@ import (
 	"github.com/charmbracelet/lipgloss"
 )
 
-func cc(hex, ansi string) lipgloss.CompleteColor {
-	return lipgloss.CompleteColor{
-		TrueColor: hex,
-		ANSI256:   hex,
-		ANSI:      ansi,
-	}
-}
-
 type Theme struct {
 	Name string
 
 	// Base layers
-	Bg      lipgloss.TerminalColor
-	Surface lipgloss.TerminalColor
-	Panel   lipgloss.TerminalColor
-	Border  lipgloss.TerminalColor
+	Bg      lipgloss.Color
+	Surface lipgloss.Color
+	Panel   lipgloss.Color
+	Border  lipgloss.Color
 
 	// Text
-	Fg     lipgloss.TerminalColor
-	Muted  lipgloss.TerminalColor
-	Subtle lipgloss.TerminalColor
+	Fg     lipgloss.Color
+	Muted  lipgloss.Color
+	Subtle lipgloss.Color
 
 	// Semantic
-	Success lipgloss.TerminalColor
-	Warning lipgloss.TerminalColor
-	Stale   lipgloss.TerminalColor
-	Danger  lipgloss.TerminalColor
-	Info    lipgloss.TerminalColor
-	Accent  lipgloss.TerminalColor
-	Purple  lipgloss.TerminalColor
+	Success lipgloss.Color
+	Warning lipgloss.Color
+	Stale   lipgloss.Color
+	Danger  lipgloss.Color
+	Info    lipgloss.Color
+	Accent  lipgloss.Color
+	Purple  lipgloss.Color
 
 	// Table
-	ZebraBg lipgloss.TerminalColor
+	ZebraBg lipgloss.Color
 
 	// Selection
-	SelectedFg lipgloss.TerminalColor
-	SelectedBg lipgloss.TerminalColor
+	SelectedFg lipgloss.Color
+	SelectedBg lipgloss.Color
 }
 
 var themes = []Theme{
 	themeFlexokiDark,
-	themeEverforest,
-	themeKanagawa,
 	themeTokyoNight,
 	themeCatppuccinMocha,
-	themeRosePine,
-	themeDracula,
+	themeNord,
+	themeGruvbox,
 }
 
 var themeFlexokiDark = Theme{
 	Name:       "Flexoki Dark",
-	Bg:         cc("#1C1B1A", ""),
-	Surface:    cc("#282726", ""),
-	Panel:      cc("#343331", ""),
-	Border:     cc("#575653", "8"),
-	Fg:         cc("#CECDC3", "15"),
-	Muted:      cc("#878580", "7"),
-	Subtle:     cc("#6F6E69", "7"),
-	Success:    cc("#879A39", "10"),
-	Warning:    cc("#D0A215", "11"),
-	Stale:      cc("#DA702C", "3"),
-	Danger:     cc("#D14D41", "9"),
-	Info:       cc("#4385BE", "12"),
-	Accent:     cc("#3AA99F", "14"),
-	Purple:     cc("#8B7EC8", "13"),
-	ZebraBg:    cc("#222120", ""),
-	SelectedFg: cc("#FFFCF0", "15"),
-	SelectedBg: cc("#403E3C", "4"),
+	Bg:         "#1C1B1A",
+	Surface:    "#282726",
+	Panel:      "#343331",
+	Border:     "#575653",
+	Fg:         "#CECDC3",
+	Muted:      "#878580",
+	Subtle:     "#6F6E69",
+	Success:    "#879A39",
+	Warning:    "#D0A215",
+	Stale:      "#DA702C",
+	Danger:     "#D14D41",
+	Info:       "#4385BE",
+	Accent:     "#3AA99F",
+	Purple:     "#8B7EC8",
+	ZebraBg:    "#222120",
+	SelectedFg: "#FFFCF0",
+	SelectedBg: "#403E3C",
 }
 
 var themeTokyoNight = Theme{
 	Name:       "Tokyo Night",
-	Bg:         cc("#1a1b26", ""),
-	Surface:    cc("#24283b", ""),
-	Panel:      cc("#292e42", ""),
-	Border:     cc("#3b4261", "8"),
-	Fg:         cc("#c0caf5", "15"),
-	Muted:      cc("#7982a9", "7"),
-	Subtle:     cc("#565f89", "7"),
-	Success:    cc("#9ece6a", "10"),
-	Warning:    cc("#e0af68", "11"),
-	Stale:      cc("#ff9e64", "3"),
-	Danger:     cc("#f7768e", "9"),
-	Info:       cc("#7aa2f7", "12"),
-	Accent:     cc("#7dcfff", "14"),
-	Purple:     cc("#bb9af7", "13"),
-	ZebraBg:    cc("#1c1d28", ""),
-	SelectedFg: cc("#c0caf5", "15"),
-	SelectedBg: cc("#363c53", "4"),
+	Bg:         "#1a1b26",
+	Surface:    "#24283b",
+	Panel:      "#292e42",
+	Border:     "#3b4261",
+	Fg:         "#c0caf5",
+	Muted:      "#a9b1d6",
+	Subtle:     "#565f89",
+	Success:    "#9ece6a",
+	Warning:    "#e0af68",
+	Stale:      "#ff9e64",
+	Danger:     "#f7768e",
+	Info:       "#7aa2f7",
+	Accent:     "#7dcfff",
+	Purple:     "#bb9af7",
+	ZebraBg:    "#1c1d28",
+	SelectedFg: "#c0caf5",
+	SelectedBg: "#292e42",
 }
 
-var themeDracula = Theme{
-	Name:       "Dracula",
-	Bg:         cc("#282a36", ""),
-	Surface:    cc("#343746", ""),
-	Panel:      cc("#44475a", ""),
-	Border:     cc("#6272a4", "8"),
-	Fg:         cc("#f8f8f2", "15"),
-	Muted:      cc("#a9b0cb", "7"),
-	Subtle:     cc("#6272a4", "7"),
-	Success:    cc("#50fa7b", "10"),
-	Warning:    cc("#f1fa8c", "11"),
-	Stale:      cc("#ffb86c", "3"),
-	Danger:     cc("#ff5555", "9"),
-	Info:       cc("#8be9fd", "12"),
-	Accent:     cc("#bd93f9", "14"),
-	Purple:     cc("#ff79c6", "13"),
-	ZebraBg:    cc("#2c2e3a", ""),
-	SelectedFg: cc("#f8f8f2", "15"),
-	SelectedBg: cc("#52556b", "4"),
+var themeGruvbox = Theme{
+	Name:       "Gruvbox",
+	Bg:         "#282828",
+	Surface:    "#3c3836",
+	Panel:      "#504945",
+	Border:     "#665c54",
+	Fg:         "#ebdbb2",
+	Muted:      "#bdae93",
+	Subtle:     "#7c6f64",
+	Success:    "#b8bb26",
+	Warning:    "#fabd2f",
+	Stale:      "#fe8019",
+	Danger:     "#fb4934",
+	Info:       "#83a598",
+	Accent:     "#8ec07c",
+	Purple:     "#d3869b",
+	ZebraBg:    "#2a2a2a",
+	SelectedFg: "#fbf1c7",
+	SelectedBg: "#504945",
 }
 
 var themeCatppuccinMocha = Theme{
 	Name:       "Catppuccin Mocha",
-	Bg:         cc("#1e1e2e", ""),
-	Surface:    cc("#313244", ""),
-	Panel:      cc("#45475a", ""),
-	Border:     cc("#585b70", "8"),
-	Fg:         cc("#cdd6f4", "15"),
-	Muted:      cc("#a6adc8", "7"),
-	Subtle:     cc("#6c7086", "7"),
-	Success:    cc("#7dc47a", "10"),
-	Warning:    cc("#f0c644", "11"),
-	Stale:      cc("#fab387", "3"),
-	Danger:     cc("#e6546e", "9"),
-	Info:       cc("#89b4fa", "12"),
-	Accent:     cc("#94e2d5", "14"),
-	Purple:     cc("#cba6f7", "13"),
-	ZebraBg:    cc("#212130", ""),
-	SelectedFg: cc("#cdd6f4", "15"),
-	SelectedBg: cc("#585b70", "4"),
+	Bg:         "#1e1e2e",
+	Surface:    "#313244",
+	Panel:      "#45475a",
+	Border:     "#585b70",
+	Fg:         "#cdd6f4",
+	Muted:      "#a6adc8",
+	Subtle:     "#6c7086",
+	Success:    "#a6e3a1",
+	Warning:    "#f9e2af",
+	Stale:      "#fab387",
+	Danger:     "#f38ba8",
+	Info:       "#89b4fa",
+	Accent:     "#94e2d5",
+	Purple:     "#cba6f7",
+	ZebraBg:    "#232334",
+	SelectedFg: "#cdd6f4",
+	SelectedBg: "#45475a",
 }
 
-var themeRosePine = Theme{
-	Name:       "Rosé Pine",
-	Bg:         cc("#191724", ""),
-	Surface:    cc("#1f1d2e", ""),
-	Panel:      cc("#26233a", ""),
-	Border:     cc("#524f67", "8"),
-	Fg:         cc("#e0def4", "15"),
-	Muted:      cc("#908caa", "7"),
-	Subtle:     cc("#6e6a86", "7"),
-	Success:    cc("#9ccfd8", "10"),
-	Warning:    cc("#f6c177", "11"),
-	Stale:      cc("#ebbcba", "3"),
-	Danger:     cc("#eb6f92", "9"),
-	Info:       cc("#3e8fb0", "12"),
-	Accent:     cc("#c4a7e7", "14"),
-	Purple:     cc("#ebbcba", "13"),
-	ZebraBg:    cc("#1c1a28", ""),
-	SelectedFg: cc("#e0def4", "15"),
-	SelectedBg: cc("#403d52", "4"),
-}
-
-var themeKanagawa = Theme{
-	Name:       "Kanagawa",
-	Bg:         cc("#1F1F28", ""),
-	Surface:    cc("#2A2A37", ""),
-	Panel:      cc("#363646", ""),
-	Border:     cc("#54546D", "8"),
-	Fg:         cc("#DCD7BA", "15"),
-	Muted:      cc("#C8C093", "7"),
-	Subtle:     cc("#727169", "7"),
-	Success:    cc("#98BB6C", "10"),
-	Warning:    cc("#E6C384", "11"),
-	Stale:      cc("#FFA066", "3"),
-	Danger:     cc("#E46876", "9"),
-	Info:       cc("#7E9CD8", "12"),
-	Accent:     cc("#7FB4CA", "14"),
-	Purple:     cc("#D27E99", "13"),
-	ZebraBg:    cc("#22222c", ""),
-	SelectedFg: cc("#DCD7BA", "15"),
-	SelectedBg: cc("#2D4F67", "4"),
-}
-
-var themeEverforest = Theme{
-	Name:       "Everforest",
-	Bg:         cc("#2F383E", ""),
-	Surface:    cc("#343F44", ""),
-	Panel:      cc("#3D484D", ""),
-	Border:     cc("#56635F", "8"),
-	Fg:         cc("#D3C6AA", "15"),
-	Muted:      cc("#9DA9A0", "7"),
-	Subtle:     cc("#7A8478", "7"),
-	Success:    cc("#A7C080", "10"),
-	Warning:    cc("#DBBC7F", "11"),
-	Stale:      cc("#E69875", "3"),
-	Danger:     cc("#E67E80", "9"),
-	Info:       cc("#7FBBB3", "12"),
-	Accent:     cc("#83C092", "14"),
-	Purple:     cc("#D699B6", "13"),
-	ZebraBg:    cc("#30393F", ""),
-	SelectedFg: cc("#D3C6AA", "15"),
-	SelectedBg: cc("#475258", "4"),
+var themeNord = Theme{
+	Name:       "Nord",
+	Bg:         "#2e3440",
+	Surface:    "#3b4252",
+	Panel:      "#434c5e",
+	Border:     "#4c566a",
+	Fg:         "#d8dee9",
+	Muted:      "#d8dee9",
+	Subtle:     "#4c566a",
+	Success:    "#a3be8c",
+	Warning:    "#ebcb8b",
+	Stale:      "#d08770",
+	Danger:     "#bf616a",
+	Info:       "#81a1c1",
+	Accent:     "#88c0d0",
+	Purple:     "#b48ead",
+	ZebraBg:    "#323845",
+	SelectedFg: "#eceff4",
+	SelectedBg: "#434c5e",
 }
 
 func (t Theme) HuhTheme() *huh.Theme {

internal/tui/tui.go

@@ -30,9 +30,9 @@ type styles struct {
 	activeTab    lipgloss.Style
 	inactiveTab  lipgloss.Style
 
-	sparkSuccess lipgloss.TerminalColor
-	sparkWarning lipgloss.TerminalColor
-	sparkDanger  lipgloss.TerminalColor
+	sparkSuccess string
+	sparkWarning string
+	sparkDanger  string
 
 	tableHeaderStyle   lipgloss.Style
 	tableCellStyle     lipgloss.Style
@@ -46,23 +46,23 @@ type styles struct {
 
 func newStyles(t Theme) *styles {
 	return &styles{
-		subtleStyle:  lipgloss.NewStyle().Foreground(t.Subtle).Faint(true),
+		subtleStyle:  lipgloss.NewStyle().Foreground(t.Subtle),
 		specialStyle: lipgloss.NewStyle().Foreground(t.Success),
-		warnStyle:    lipgloss.NewStyle().Foreground(t.Warning).Bold(true),
-		staleStyle:   lipgloss.NewStyle().Foreground(t.Stale).Faint(true),
-		dangerStyle:  lipgloss.NewStyle().Foreground(t.Danger).Bold(true),
+		warnStyle:    lipgloss.NewStyle().Foreground(t.Warning),
+		staleStyle:   lipgloss.NewStyle().Foreground(t.Stale),
+		dangerStyle:  lipgloss.NewStyle().Foreground(t.Danger),
 		titleStyle:   lipgloss.NewStyle().Foreground(t.Accent).Bold(true),
 		activeTab:    lipgloss.NewStyle().Background(t.Surface).Foreground(t.Accent).Bold(true).Padding(0, 1),
-		inactiveTab:  lipgloss.NewStyle().Padding(0, 1).Foreground(t.Muted).Faint(true),
+		inactiveTab:  lipgloss.NewStyle().Padding(0, 1).Foreground(t.Muted),
 
-		sparkSuccess: t.Success,
-		sparkWarning: t.Warning,
-		sparkDanger:  t.Danger,
+		sparkSuccess: string(t.Success),
+		sparkWarning: string(t.Warning),
+		sparkDanger:  string(t.Danger),
 
 		tableHeaderStyle:   lipgloss.NewStyle().Foreground(t.Accent).Bold(true).Padding(0, 1),
 		tableCellStyle:     lipgloss.NewStyle().Padding(0, 1),
 		tableSelectedStyle: lipgloss.NewStyle().Padding(0, 1).Bold(true).Foreground(t.SelectedFg).Background(t.SelectedBg),
-		tableBorderStyle:   lipgloss.NewStyle().Foreground(t.Border).Faint(true),
+		tableBorderStyle:   lipgloss.NewStyle().Foreground(t.Border),
 		tableZebraStyle:    lipgloss.NewStyle().Padding(0, 1).Background(t.ZebraBg),
 
 		siteGroupStyle: lipgloss.NewStyle().Padding(0, 1).Bold(true).Foreground(t.Accent),
@@ -80,82 +80,41 @@ const (
 	chromeFooter = 2 // footer: "\n" prefix + text line
 	chromeTable  = 3 // renderTable "\n" prefix + top border + header + bottom border (lipgloss collapses two into three rendered lines)
 	chromeBase   = chromePadV + chromeHeader + chromeGaps + chromeFooter + chromeTable
-)
 
-const (
-	sectionAlerts = 0
-	sectionNodes  = 1
-	sectionUsers  = 2
-)
-
-const (
-	panelMonitors = 0
-	panelLogs     = 1
-	panelDetail   = 2
-	panelMaint    = 3
-)
-
-type bottomPanel int
-
-const (
-	bottomNone bottomPanel = iota
-	bottomLogs
-	bottomMaint
-)
-
-const (
-	detailDefault = 0
-	detailSLA     = 1
-	detailHistory = 2
-)
-
-const (
-	sortStatus  = 0
-	sortName    = 1
-	sortLatency = 2
-	sortMax     = 3
+	detailSparkWidth = 40
 )
 
 type sessionState int
 
 const (
-	stateDashboard        sessionState = 0
-	stateLogs             sessionState = 1
-	stateDetailFullscreen sessionState = 2
-	stateAlertDetail      sessionState = 3
-	stateFormSite         sessionState = 4
-	stateFormAlert        sessionState = 5
-	stateFormUser         sessionState = 6
-	stateConfirmDelete    sessionState = 7
-	stateFormMaint        sessionState = 8
-	stateSettings         sessionState = 11
-	stateMaintDetail      sessionState = 12
+	stateDashboard sessionState = iota
+	stateLogs
+	stateUsers
+	stateDetail
+	stateAlertDetail
+	stateFormSite
+	stateFormAlert
+	stateFormUser
+	stateConfirmDelete
+	stateFormMaint
+	stateHistory
+	stateSLA
 )
 
 type Model struct {
-	state           sessionState
-	returnState     sessionState
-	settingsSection int
-	settingsCursor  int
-	settingsOffset  int
-	cursor          int
-	selectedID      int
-	sortColumn      int
-	sortAsc         bool
-	tableOffset     int
-	maxTableRows    int
-	termWidth       int
-	termHeight      int
-	contentWidth    int
-	focusedPanel    int
-	detailMode      int
-	logScrollOffset int
-	editID          int
-	editToken       string
+	state        sessionState
+	currentTab   int
+	cursor       int
+	selectedID   int
+	tableOffset  int
+	maxTableRows int
+	termWidth    int
+	termHeight   int
+	editID       int
+	editToken    string
 
 	huhForm       *huh.Form
 	siteFormData  *siteFormData
-	lastSiteType  string
 	alertFormData *alertFormData
 	userFormData  *userFormData
 	maintFormData *maintFormData
@@ -165,27 +124,25 @@ type Model struct {
 	logTotal           int
 	logShown           int
 
+	historyViewport viewport.Model
 	historyChanges  []models.StateChange
 	historySiteName string
 	historySiteID   int
 
-	slaReport          monitor.SLAReport
-	slaDailyBreakdown  []monitor.DayReport
-	slaSiteName        string
-	slaSiteID          int
-	slaPeriodIdx       int
-	detailScrollOffset int
+	slaViewport       viewport.Model
+	slaReport         monitor.SLAReport
+	slaDailyBreakdown []monitor.DayReport
+	slaSiteName       string
+	slaSiteID         int
+	slaPeriodIdx      int
 
 	isAdmin bool
 	zones   *zone.Manager
 
 	deleteID   int
 	deleteName string
-	deleteKind string
+	deleteTab  int
 
-	maintDetailID int
-
-	ctx        context.Context
 	collapsed  map[int]bool
 	store      store.Store
 	engine     *monitor.Engine
@@ -207,18 +164,14 @@ type Model struct {
 	lastTabLoad        time.Time // last dispatch of loadTabDataCmd (throttle)
 	tabSeq             int       // seq of the newest issued tab-data load
 
-	maintSet            map[int]bool
-	bottomPanel         bottomPanel
-	detailOpen          bool
-	maintCursor         int
+	// detail-panel state-change history, loaded on enter so View does no DB IO
 	detailChanges       []models.StateChange
 	detailChangesSiteID int
-	detailDailyDays     []monitor.DayReport
 
 	filterMode bool
 	filterText string
 
-	pendingG bool
+	sparkTooltipIdx int // clicked sparkline data index, -1 = none
 
 	// demoMode renders a stable status dot instead of the animated pulse so
 	// screenshots/recordings don't capture the spinner mid-frame. Set via UPTOP_DEMO=1.
@@ -226,14 +179,14 @@ type Model struct {
 	version  string
 }
 
-func InitialModel(ctx context.Context, isAdmin bool, s store.Store, eng *monitor.Engine, version string) Model {
+func InitialModel(isAdmin bool, s store.Store, eng *monitor.Engine, version string) Model {
 	vpLogs := viewport.New(100, 20)
 	vpLogs.SetContent("Waiting for logs...")
 	z := zone.New()
 	spring := harmonica.NewSpring(harmonica.FPS(10), 6.0, 0.4)
-	collapsed := loadCollapsed(ctx, s)
+	collapsed := loadCollapsed(s)
 
-	themeName, _ := s.GetPreference(ctx, "theme")
+	themeName, _ := s.GetPreference(context.Background(), "theme")
 	theme := themeByName(themeName)
 	themeIdx := 0
 	for i, t := range themes {
@@ -243,36 +196,22 @@ func InitialModel(ctx context.Context, isAdmin bool, s store.Store, eng *monitor
 		}
 	}
 
-	detailPref, _ := s.GetPreference(ctx, "detail_open")
-
-	bp := bottomLogs
-	if bpPref, _ := s.GetPreference(ctx, "bottom_panel"); bpPref != "" {
-		switch bpPref {
-		case "none":
-			bp = bottomNone
-		case "maint":
-			bp = bottomMaint
-		}
-	}
-
 	return Model{
-		ctx:          ctx,
-		state:        stateDashboard,
-		logViewport:  vpLogs,
-		maxTableRows: 5,
-		isAdmin:      isAdmin,
-		store:        s,
-		engine:       eng,
-		zones:        z,
-		pulseSpring:  spring,
-		collapsed:    collapsed,
-		theme:        theme,
-		themeIndex:   themeIdx,
-		st:           newStyles(theme),
-		bottomPanel:  bp,
-		detailOpen:   detailPref == "true",
-		demoMode:     os.Getenv("UPTOP_DEMO") == "1",
-		version:      version,
+		state:           stateDashboard,
+		logViewport:     vpLogs,
+		maxTableRows:    5,
+		isAdmin:         isAdmin,
+		store:           s,
+		engine:          eng,
+		zones:           z,
+		pulseSpring:     spring,
+		collapsed:       collapsed,
+		theme:           theme,
+		themeIndex:      themeIdx,
+		st:              newStyles(theme),
+		demoMode:        os.Getenv("UPTOP_DEMO") == "1",
+		version:         version,
+		sparkTooltipIdx: -1,
 	}
 }
 

internal/tui/update_test.go

@@ -118,10 +118,11 @@ func TestDetailLoad_CachesAndViewDoesNoIO(t *testing.T) {
 	m := newTestModel(ms)
 	m.sites = []models.Site{{SiteConfig: models.SiteConfig{ID: 1, Name: "site"}, SiteState: models.SiteState{Status: "DOWN"}}}
 	m.cursor = 0
-	m.detailOpen = true
+	m.state = stateDetail
 	m.termWidth = 120
 	m.termHeight = 40
 
+	// Entering detail dispatches the load Cmd.
 	cmd := m.loadDetailCmd(1)
 	if cmd == nil {
 		t.Fatal("loadDetailCmd returned nil")
@@ -135,14 +136,16 @@ func TestDetailLoad_CachesAndViewDoesNoIO(t *testing.T) {
 		t.Fatalf("expected exactly 1 store hit from the load Cmd, got %d", ms.stateChangeCalls)
 	}
 
+	// Apply the msg through Update (caches into the model).
 	updated, _ := m.Update(dd)
 	m = updated.(Model)
 	if m.detailChangesSiteID != 1 || len(m.detailChanges) != 1 {
 		t.Fatalf("detail changes not cached: id=%d n=%d", m.detailChangesSiteID, len(m.detailChanges))
 	}
 
+	// Render the detail panel several times — it must read the cache, not the DB.
 	for i := 0; i < 3; i++ {
-		_ = m.viewDetailInline(80, 30)
+		_ = m.viewDetailPanel()
 	}
 	if ms.stateChangeCalls != 1 {
 		t.Errorf("View performed DB IO: store hit %d times (want 1, from the Cmd only)", ms.stateChangeCalls)
@@ -199,16 +202,16 @@ func TestHistoryKey_LoadsOffUIGoroutine(t *testing.T) {
 	ms := &tuiMockStore{stateChanges: []models.StateChange{{FromStatus: "UP", ToStatus: "DOWN"}}}
 	m := newTestModel(ms)
 	m.sites = []models.Site{{SiteConfig: models.SiteConfig{ID: 7, Name: "site"}}}
-	m.detailOpen = true
+	m.state = stateDetail
 	m.termWidth, m.termHeight = 120, 40
 
-	updated, cmd := (&m).handleDashboardKey(keyMsg("h"))
+	updated, cmd := (&m).handleDetailKey(keyMsg("h"))
 	if ms.stateChangeCalls != 0 {
 		t.Fatal("history keypress hit the store synchronously in Update")
 	}
 	got := updated.(*Model)
-	if got.detailMode != detailHistory || got.historySiteID != 7 {
-		t.Fatalf("history mode not set: mode=%v siteID=%d", got.detailMode, got.historySiteID)
+	if got.state != stateHistory || got.historySiteID != 7 {
+		t.Fatalf("history view not opened: state=%v siteID=%d", got.state, got.historySiteID)
 	}
 	if cmd == nil {
 		t.Fatal("expected a history load Cmd")
@@ -226,6 +229,7 @@ func TestHistoryKey_LoadsOffUIGoroutine(t *testing.T) {
 		t.Fatal("history reply not folded into the model")
 	}
 
+	// A reply for a previously opened site must not clobber the current one.
 	m2.historySiteID = 9
 	stale, _ := m2.Update(historyDataMsg{siteID: 7, changes: nil})
 	if m3 := stale.(Model); len(m3.historyChanges) != 1 {
@@ -237,15 +241,20 @@ func TestSLAData_DropsStaleReply(t *testing.T) {
 	m := newTestModel(&tuiMockStore{})
 	m.termWidth, m.termHeight = 120, 40
 	m.sites = []models.Site{{SiteConfig: models.SiteConfig{ID: 3}, SiteState: models.SiteState{Status: "UP"}}}
-	m.detailOpen = true
-	m.slaSiteID = 3
-	m.slaPeriodIdx = 2
 
+	if cmd := (&m).openSLAView(m.sites[0]); cmd == nil {
+		t.Fatal("openSLAView should return a load Cmd")
+	}
+
+	// Reply for a different period than currently selected → dropped.
+	// (slaDataMsg routes through a pointer-receiver handler, so Update
+	// returns *Model on this path.)
 	updated, _ := m.Update(slaDataMsg{siteID: 3, periodIdx: 0})
 	if mm := updated.(*Model); mm.slaDailyBreakdown != nil {
 		t.Error("stale SLA reply (old period) was applied")
 	}
 
+	// Matching reply → report computed.
 	updated, _ = updated.(*Model).Update(slaDataMsg{siteID: 3, periodIdx: m.slaPeriodIdx})
 	if mm := updated.(*Model); mm.slaDailyBreakdown == nil {
 		t.Error("matching SLA reply was not applied")
@@ -257,7 +266,7 @@ func TestConfirmDelete_WritesOffUIGoroutine(t *testing.T) {
 	m := newTestModel(ms)
 	m.sites = []models.Site{{SiteConfig: models.SiteConfig{ID: 4, Name: "s"}}}
 	m.state = stateConfirmDelete
-	m.deleteKind = "site"
+	m.deleteTab = 0
 	m.deleteID = 4
 
 	updated, cmd := (&m).handleConfirmDelete(keyMsg("y"))
@@ -291,7 +300,7 @@ func TestWriteDoneMsg_LogsErrorAndReloads(t *testing.T) {
 	mm := updated.(Model)
 	found := false
 	for _, line := range mm.engine.GetLogs() {
-		if strings.Contains(line.Message, "Delete site failed: boom") {
+		if strings.Contains(line, "Delete site failed: boom") {
 			found = true
 		}
 	}
@@ -300,81 +309,17 @@ func TestWriteDoneMsg_LogsErrorAndReloads(t *testing.T) {
 	}
 }
 
-func TestEnterNarrowTerminal_OpensFullscreen(t *testing.T) {
-	ms := &tuiMockStore{stateChanges: []models.StateChange{{FromStatus: "UP", ToStatus: "DOWN"}}}
-	m := newTestModel(ms)
-	m.sites = []models.Site{{SiteConfig: models.SiteConfig{ID: 1, Name: "site"}, SiteState: models.SiteState{Status: "UP"}}}
-	m.termWidth = 80 // narrow, below wideBreakpoint (120)
-	m.termHeight = 40
-	m.focusedPanel = panelMonitors
-
-	updated, cmd := (&m).handleDashboardKey(keyMsg("enter"))
-	got := updated.(*Model)
-	if got.state != stateDetailFullscreen {
-		t.Fatalf("expected stateDetailFullscreen, got %d", got.state)
-	}
-	if cmd == nil {
-		t.Fatal("expected a detail load Cmd")
-	}
-}
-
-func TestEnterWideTerminal_TogglesSidebar(t *testing.T) {
-	ms := &tuiMockStore{stateChanges: []models.StateChange{{FromStatus: "UP", ToStatus: "DOWN"}}}
-	m := newTestModel(ms)
-	m.sites = []models.Site{{SiteConfig: models.SiteConfig{ID: 1, Name: "site"}, SiteState: models.SiteState{Status: "UP"}}}
-	m.termWidth = 140 // wide, above wideBreakpoint (120)
-	m.termHeight = 40
-	m.focusedPanel = panelMonitors
-
-	updated, _ := (&m).handleDashboardKey(keyMsg("enter"))
-	got := updated.(*Model)
-	if got.state != stateDashboard {
-		t.Fatalf("expected stateDashboard, got %d", got.state)
-	}
-	if !got.detailOpen {
-		t.Fatal("expected detailOpen to be true")
-	}
-}
-
-func TestDetailFullscreen_EscReturnsToDashboard(t *testing.T) {
-	m := newTestModel(&tuiMockStore{})
-	m.state = stateDetailFullscreen
-	m.sites = []models.Site{{SiteConfig: models.SiteConfig{ID: 1, Name: "site"}}}
-
-	updated, _ := (&m).handleDetailFullscreen(tea.KeyMsg{Type: tea.KeyEsc})
-	got := updated.(*Model)
-	if got.state != stateDashboard {
-		t.Fatalf("expected stateDashboard after Esc, got %d", got.state)
-	}
-	if got.focusedPanel != panelMonitors {
-		t.Fatalf("expected panelMonitors focus, got %d", got.focusedPanel)
-	}
-}
-
-func TestDetailRefreshCmd_FiresForFullscreen(t *testing.T) {
-	ms := &tuiMockStore{stateChanges: []models.StateChange{{FromStatus: "UP", ToStatus: "DOWN"}}}
-	m := newTestModel(ms)
-	m.sites = []models.Site{{SiteConfig: models.SiteConfig{ID: 5, Name: "site"}}}
-	m.state = stateDetailFullscreen
-	m.detailOpen = false
-
-	cmd := (&m).detailRefreshCmd()
-	if cmd == nil {
-		t.Fatal("detailRefreshCmd should fire for stateDetailFullscreen")
-	}
-}
-
 func TestDetailRefreshCmd_OnlyWhileDetailOpen(t *testing.T) {
 	ms := &tuiMockStore{stateChanges: []models.StateChange{{FromStatus: "UP", ToStatus: "DOWN"}}}
 	m := newTestModel(ms)
 	m.sites = []models.Site{{SiteConfig: models.SiteConfig{ID: 5, Name: "site"}}}
 
-	m.detailOpen = false
+	m.state = stateDashboard
 	if (&m).detailRefreshCmd() != nil {
 		t.Error("refresh Cmd issued outside the detail view")
 	}
 
-	m.detailOpen = true
+	m.state = stateDetail
 	cmd := (&m).detailRefreshCmd()
 	if cmd == nil {
 		t.Fatal("open detail panel should refresh on the tab-data cadence")
@@ -389,86 +334,3 @@ func TestDetailRefreshCmd_OnlyWhileDetailOpen(t *testing.T) {
 		t.Error("refresh Cmd issued for an out-of-range cursor")
 	}
 }
-
-func TestBuildMaintSet_GlobalWindow(t *testing.T) {
-	m := newTestModel(&tuiMockStore{})
-	m.sites = []models.Site{
-		{SiteConfig: models.SiteConfig{ID: 1, Name: "a"}},
-		{SiteConfig: models.SiteConfig{ID: 2, Name: "b"}},
-	}
-	m.maintenanceWindows = []models.MaintenanceWindow{
-		{ID: 1, MonitorID: 0, Type: "maintenance", StartTime: time.Now().Add(-time.Hour)},
-	}
-	m.buildMaintSet()
-	if !m.maintSet[1] || !m.maintSet[2] {
-		t.Error("global maint window should mark all monitors")
-	}
-}
-
-func TestBuildMaintSet_TargetedWindow(t *testing.T) {
-	m := newTestModel(&tuiMockStore{})
-	m.sites = []models.Site{
-		{SiteConfig: models.SiteConfig{ID: 1, Name: "a"}},
-		{SiteConfig: models.SiteConfig{ID: 2, Name: "b"}},
-	}
-	m.maintenanceWindows = []models.MaintenanceWindow{
-		{ID: 1, MonitorID: 1, Type: "maintenance", StartTime: time.Now().Add(-time.Hour)},
-	}
-	m.buildMaintSet()
-	if !m.maintSet[1] {
-		t.Error("targeted window should mark monitor 1")
-	}
-	if m.maintSet[2] {
-		t.Error("targeted window should NOT mark monitor 2")
-	}
-}
-
-func TestBuildMaintSet_GroupPropagates(t *testing.T) {
-	m := newTestModel(&tuiMockStore{})
-	m.sites = []models.Site{
-		{SiteConfig: models.SiteConfig{ID: 10, Name: "group", Type: "group"}},
-		{SiteConfig: models.SiteConfig{ID: 11, Name: "child1", ParentID: 10}},
-		{SiteConfig: models.SiteConfig{ID: 12, Name: "child2", ParentID: 10}},
-		{SiteConfig: models.SiteConfig{ID: 20, Name: "other"}},
-	}
-	m.maintenanceWindows = []models.MaintenanceWindow{
-		{ID: 1, MonitorID: 10, Type: "maintenance", StartTime: time.Now().Add(-time.Hour)},
-	}
-	m.buildMaintSet()
-	if !m.maintSet[10] || !m.maintSet[11] || !m.maintSet[12] {
-		t.Error("group maint window should mark group + children")
-	}
-	if m.maintSet[20] {
-		t.Error("unrelated monitor should NOT be marked")
-	}
-}
-
-func TestBuildMaintSet_ExpiredIgnored(t *testing.T) {
-	m := newTestModel(&tuiMockStore{})
-	m.sites = []models.Site{
-		{SiteConfig: models.SiteConfig{ID: 1, Name: "a"}},
-	}
-	m.maintenanceWindows = []models.MaintenanceWindow{
-		{ID: 1, MonitorID: 1, Type: "maintenance",
-			StartTime: time.Now().Add(-2 * time.Hour),
-			EndTime:   time.Now().Add(-1 * time.Hour)},
-	}
-	m.buildMaintSet()
-	if m.maintSet[1] {
-		t.Error("expired maint window should not mark monitor")
-	}
-}
-
-func TestBuildMaintSet_IncidentIgnored(t *testing.T) {
-	m := newTestModel(&tuiMockStore{})
-	m.sites = []models.Site{
-		{SiteConfig: models.SiteConfig{ID: 1, Name: "a"}},
-	}
-	m.maintenanceWindows = []models.MaintenanceWindow{
-		{ID: 1, MonitorID: 0, Type: "incident", StartTime: time.Now().Add(-time.Hour)},
-	}
-	m.buildMaintSet()
-	if m.maintSet[1] {
-		t.Error("incident windows should not mark monitors as in maintenance")
-	}
-}

internal/tui/uptime_timeline.go

@@ -1,50 +0,0 @@
-package tui
-
-import (
-	"fmt"
-	"strings"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/monitor"
-)
-
-func (m Model) uptimeTimeline(days []monitor.DayReport, width int) string {
-	if len(days) == 0 {
-		return m.st.subtleStyle.Render("No uptime data")
-	}
-
-	maxDays := width - 10
-	if maxDays < 10 {
-		maxDays = 10
-	}
-
-	display := days
-	if len(display) > maxDays {
-		display = display[len(display)-maxDays:]
-	}
-
-	var sb strings.Builder
-	for _, d := range display {
-		ch := "█"
-		switch {
-		case d.UptimePct >= uptimeGoodPct:
-			sb.WriteString(m.st.specialStyle.Render(ch))
-		case d.UptimePct >= uptimeWarnPct:
-			sb.WriteString(m.st.warnStyle.Render(ch))
-		case d.UptimePct > 0:
-			sb.WriteString(m.st.dangerStyle.Render(ch))
-		default:
-			sb.WriteString(m.st.subtleStyle.Render("░"))
-		}
-	}
-
-	pct := days[len(days)-1].UptimePct
-	pctStyle := m.st.specialStyle
-	if pct < uptimeGoodPct {
-		pctStyle = m.st.dangerStyle
-	} else if pct < uptimeExcellentPct {
-		pctStyle = m.st.warnStyle
-	}
-	sb.WriteString("  " + pctStyle.Render(fmt.Sprintf("%.2f%%", pct)))
-
-	return sb.String()
-}

internal/tui/view_dashboard.go

@@ -47,12 +47,12 @@ func (m Model) View() string {
 	switch m.state {
 	case stateConfirmDelete:
 		kind := "monitor"
-		switch m.deleteKind {
-		case "maint":
-			kind = "maintenance window"
-		case "alert":
+		switch m.deleteTab {
+		case 1:
 			kind = "alert"
-		case "user":
+		case 4:
+			kind = "maintenance window"
+		case 5:
 			kind = "user"
 		}
 		msg := m.st.dangerStyle.Render(fmt.Sprintf("Delete %s \"%s\"?", kind, m.deleteName))
@@ -90,16 +90,14 @@ func (m Model) View() string {
 			return lipgloss.NewStyle().Padding(1, 2).Render(header + "\n\n" + m.huhForm.View() + "\n" + footer)
 		}
 		return ""
-	case stateLogs:
-		return m.viewLogsFullscreen()
-	case stateDetailFullscreen:
-		return m.viewDetailFullscreen()
+	case stateDetail:
+		return m.zones.Scan(m.viewDetailPanel())
+	case stateHistory:
+		return m.viewHistoryPanel()
+	case stateSLA:
+		return m.viewSLAPanel()
 	case stateAlertDetail:
 		return m.viewAlertDetailPanel()
-	case stateSettings:
-		return m.viewSettingsOverlay()
-	case stateMaintDetail:
-		return m.viewMaintDetailPanel()
 	default:
 		return m.zones.Scan(m.viewDashboard())
 	}
@@ -132,7 +130,7 @@ func (m Model) computeStats() dashboardStats {
 		}
 	}
 	for _, n := range m.nodes {
-		if !n.LastSeen.IsZero() && time.Since(n.LastSeen) > nodeStaleThreshold {
+		if !n.LastSeen.IsZero() && time.Since(n.LastSeen) > 5*time.Minute {
 			s.offlineNodes++
 		}
 	}
@@ -145,71 +143,29 @@ func (m Model) computeStats() dashboardStats {
 	return s
 }
 
-func (m Model) viewMonitorsLayout() string {
-	availW := m.termWidth - chromePadH
-	wide := m.termWidth >= wideBreakpoint
-
-	showDetail := m.detailOpen && wide
-
-	var detailW, monW int
-	if showDetail {
-		monW = availW * 60 / 100
-		detailW = availW - monW
-	} else {
-		monW = availW
-	}
-
-	m.contentWidth = monW - 2
-
-	monTargetH := m.maxTableRows + 5
-	monitors := m.viewSitesTab()
-	monPanel := m.zones.Mark("panel-monitors", m.titledPanelH("Monitors", monitors, "", monW, monTargetH, 0, scrollbar{pos: m.tableOffset, total: len(m.sites), visible: m.maxTableRows}, m.focusedPanel == panelMonitors))
-
-	var topParts []string
-	topParts = append(topParts, monPanel)
-	if showDetail {
-		title := ""
-		if m.cursor < len(m.sites) {
-			title = m.sites[m.cursor].Name
-		}
-		switch m.detailMode {
-		case detailSLA:
-			title = "SLA · " + title
-		case detailHistory:
-			title = "History · " + title
-		}
-		monHeight := lipgloss.Height(monPanel)
-		detail := m.viewDetailInline(detailW-2, monHeight)
-		footer := m.detailFooter(detailW - 2)
-		detailPanel := m.zones.Mark("panel-detail", m.titledPanelH(title, detail, footer, detailW, monHeight, m.detailScrollOffset, scrollbar{}, m.focusedPanel == panelDetail))
-		topParts = append(topParts, detailPanel)
-	}
-
-	top := lipgloss.JoinHorizontal(lipgloss.Top, topParts...)
-
-	switch m.bottomPanel {
-	case bottomLogs:
-		maxLines := logsStripHeight - 2
-		logContent := m.viewLogsStrip(availW-2, maxLines)
-		totalLogs := m.filteredLogCount()
-		logPanel := m.zones.Mark("panel-logs", m.titledPanelH("Logs", logContent, "", availW, logsStripHeight, 0, scrollbar{pos: m.logScrollOffset, total: totalLogs, visible: maxLines}, m.focusedPanel == panelLogs))
-		return top + "\n" + logPanel
-	case bottomMaint:
-		maxLines := logsStripHeight - 2
-		maintContent := m.viewMaintStrip(availW-2, maxLines)
-		totalMaint := len(m.activeMaintWindows())
-		maintPanel := m.zones.Mark("panel-maint", m.titledPanelH("Maint", maintContent, "", availW, logsStripHeight, 0, scrollbar{pos: 0, total: totalMaint, visible: maxLines}, m.focusedPanel == panelMaint))
-		return top + "\n" + maintPanel
-	}
-	return top
-}
-
 func (m Model) viewDashboard() string {
 	stats := m.computeStats()
 
-	header := m.renderStatusLine(stats)
+	header := m.renderTabBar(stats)
+	header = m.pulseIndicator() + " " + header
 
-	content := m.viewMonitorsLayout()
+	var content string
+	switch m.currentTab {
+	case 0:
+		content = m.viewSitesTab()
+	case 1:
+		content = m.viewAlertsTab()
+	case 2:
+		content = m.viewLogsTab()
+	case 3:
+		content = m.viewNodesTab()
+	case 4:
+		content = m.viewMaintTab()
+	case 5:
+		if m.isAdmin {
+			content = m.viewUsersTab()
+		}
+	}
 
 	content = strings.TrimSpace(content)
 	footer := m.renderFooter(stats)
@@ -230,8 +186,53 @@ func (m Model) viewDashboard() string {
 	return outerPad.Render(lipgloss.JoinVertical(lipgloss.Top, header, paddedContent, footer))
 }
 
-func (m Model) renderStatusLine(stats dashboardStats) string {
-	dot := m.st.subtleStyle.Render(" · ")
+type tabEntry struct {
+	name  string
+	count int
+	warn  int
+}
+
+func (m Model) renderTabBar(stats dashboardStats) string {
+	tabs := []tabEntry{
+		{"Sites", stats.totalMonitors, stats.downCount + stats.lateCount},
+		{"Alerts", len(m.alerts), 0},
+		{"Logs", 0, 0},
+		{"Nodes", len(m.nodes), stats.offlineNodes},
+		{"Maint", len(m.maintenanceWindows), stats.activeMaint},
+	}
+	if m.isAdmin {
+		tabs = append(tabs, tabEntry{"Users", len(m.users), 0})
+	}
+
+	countStyle := lipgloss.NewStyle().Foreground(m.theme.Muted)
+
+	var renderedTabs []string
+	for i, t := range tabs {
+		label := t.name
+		if t.count > 0 {
+			badge := countStyle.Render(fmt.Sprintf(" %d", t.count))
+			if t.warn > 0 {
+				badge = m.st.dangerStyle.Render(fmt.Sprintf(" %d", t.warn))
+			}
+			label += badge
+		}
+
+		var rendered string
+		if i == m.currentTab {
+			rendered = m.st.activeTab.Render(label)
+		} else {
+			rendered = m.st.inactiveTab.Render(label)
+		}
+		renderedTabs = append(renderedTabs, m.zones.Mark(fmt.Sprintf("tab-%d", i), rendered))
+	}
+	return lipgloss.JoinHorizontal(lipgloss.Top, renderedTabs...)
+}
+
+func (m Model) renderFooter(stats dashboardStats) string {
+	if m.filterMode {
+		cursor := lipgloss.NewStyle().Foreground(m.theme.Accent).Render("│")
+		return "\n" + m.st.titleStyle.Render("/") + " " + m.filterText + cursor + "  " + m.st.subtleStyle.Render("[Enter]Apply [Esc]Clear")
+	}
 
 	upCount := stats.totalMonitors - stats.downCount - stats.lateCount
 	var upStr string
@@ -242,83 +243,45 @@ func (m Model) renderStatusLine(stats dashboardStats) string {
 	} else {
 		upStr = m.st.specialStyle.Render(fmt.Sprintf("%d/%d UP", upCount, stats.totalMonitors))
 	}
-
-	parts := []string{m.pulseIndicator() + " " + upStr}
-
+	statusParts := []string{upStr}
 	if stats.lateCount > 0 {
-		parts = append(parts, m.st.warnStyle.Render(fmt.Sprintf("%d late", stats.lateCount)))
-	}
-	if stats.activeMaint > 0 {
-		parts = append(parts, m.st.maintStyle.Render(fmt.Sprintf("%d maint", stats.activeMaint)))
+		statusParts = append(statusParts, m.st.warnStyle.Render(fmt.Sprintf("%d LATE", stats.lateCount)))
 	}
 	if len(m.nodes) > 0 {
 		online := 0
 		for _, n := range m.nodes {
-			if !n.LastSeen.IsZero() && time.Since(n.LastSeen) < nodeOnlineThreshold {
+			if !n.LastSeen.IsZero() && time.Since(n.LastSeen) < 60*time.Second {
 				online++
 			}
 		}
-		label := "probes"
+		probeLabel := "probes"
 		if online == 1 {
-			label = "probe"
+			probeLabel = "probe"
 		}
-		parts = append(parts, fmt.Sprintf("%d %s", online, label))
+		statusParts = append(statusParts, fmt.Sprintf("%d %s", online, probeLabel))
+	}
+	statusLine := strings.Join(statusParts, m.st.subtleStyle.Render(" · "))
+
+	var keys string
+	switch m.currentTab {
+	case 0:
+		keys = "[/]Filter [n]New [e]Edit [i]Info [d]Del [p]Pause [Space]Collapse [T]Theme [Tab]Switch [q]Quit"
+	case 1:
+		keys = "[n]New [e]Edit [i]Info [d]Del [t]Test [T]Theme [Tab]Switch [q]Quit"
+	case 2:
+		keys = "[↑/↓]Scroll  [PgUp/PgDn]Page  [f]Filter  [T]Theme  [Tab]Switch  [q]Quit"
+	case 4:
+		keys = "[n]New [x]End [d]Del [T]Theme [Tab]Switch [q]Quit"
+	case 5:
+		keys = "[n]Add [d]Revoke [T]Theme [Tab]Switch [q]Quit"
+	default:
+		keys = "[T]Theme [Tab]Switch [q]Quit"
 	}
 
-	left := strings.Join(parts, dot)
 	ver := m.st.subtleStyle.Render("v" + m.version)
-
-	padW := m.termWidth - chromePadH - lipgloss.Width(left) - lipgloss.Width(ver)
-	if padW < 2 {
-		padW = 2
+	footer := statusLine + "  " + m.st.subtleStyle.Render(keys) + "  " + ver
+	if m.filterText != "" && m.currentTab == 0 {
+		footer = m.st.subtleStyle.Render(fmt.Sprintf("filter: %s", m.filterText)) + "  " + statusLine + "  " + m.st.subtleStyle.Render(keys) + "  " + ver
 	}
-
-	return left + strings.Repeat(" ", padW) + ver
-}
-
-func (m Model) hotkey(key, desc string) string {
-	k := lipgloss.NewStyle().Foreground(m.theme.Accent).Render(key)
-	d := m.st.subtleStyle.Render(desc)
-	return k + " " + d
-}
-
-func (m Model) renderFooter(_ dashboardStats) string {
-	dot := m.st.subtleStyle.Render(" · ")
-
-	if m.filterMode {
-		cursor := lipgloss.NewStyle().Foreground(m.theme.Accent).Render("│")
-		keys := m.hotkey("Enter", "Apply") + dot + m.hotkey("Esc", "Clear")
-		return "\n" + m.st.titleStyle.Render("/") + " " + m.filterText + cursor + "  " + keys
-	}
-
-	var parts []string
-	if m.focusedPanel == panelMaint {
-		parts = []string{m.hotkey("n", "New"), m.hotkey("Enter", "Detail"), m.hotkey("x", "End"), m.hotkey("d", "Del"), m.hotkey("m/Esc", "Back")}
-	} else if m.focusedPanel == panelLogs {
-		parts = []string{m.hotkey("↑/↓", "Scroll"), m.hotkey("Enter", "Expand"), m.hotkey("l/Esc", "Back")}
-	} else if m.detailOpen && m.detailMode == detailSLA {
-		parts = []string{m.hotkey("1-4", "Period"), m.hotkey("Esc", "Back")}
-	} else if m.detailOpen && m.detailMode == detailHistory {
-		parts = []string{m.hotkey("Esc", "Back")}
-	} else if m.detailOpen {
-		parts = []string{m.hotkey("Enter", "Close"), m.hotkey("h", "History"), m.hotkey("s", "SLA"), m.hotkey("e", "Edit")}
-	} else {
-		parts = []string{m.hotkey("/", "Filter"), m.hotkey("Enter", "Detail")}
-		if m.cursor < len(m.sites) && m.sites[m.cursor].Type == "group" {
-			if m.collapsed[m.sites[m.cursor].ID] {
-				parts = append(parts, m.hotkey("Space", "Expand"))
-			} else {
-				parts = append(parts, m.hotkey("Space", "Collapse"))
-			}
-		}
-		parts = append(parts, m.hotkey("n", "New"), m.hotkey("e", "Edit"), m.hotkey("d", "Del"))
-	}
-	parts = append(parts, m.hotkey("m", "Maint"), m.hotkey("l", "Logs"), m.hotkey("S", "Settings"), m.hotkey("T", "Theme"), m.hotkey("q", "Quit"))
-
-	line := strings.Join(parts, dot)
-	if m.filterText != "" {
-		line = m.st.subtleStyle.Render(fmt.Sprintf("filter: %s  ", m.filterText)) + line
-	}
-
-	return "\n" + line
+	return footer
 }

internal/tui/view_detail.go

@@ -0,0 +1,306 @@
+package tui
+
+import (
+	"fmt"
+	"sort"
+	"strconv"
+	"strings"
+	"time"
+
+	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
+	"gitea.lerkolabs.com/lerkolabs/uptop/internal/monitor"
+	"github.com/charmbracelet/lipgloss"
+)
+
+func (m Model) viewDetailPanel() string {
+	if m.cursor >= len(m.sites) {
+		return ""
+	}
+	site := m.sites[m.cursor]
+	hist, _ := m.engine.GetHistory(site.ID)
+
+	var b strings.Builder
+
+	var breadcrumb string
+	if site.ParentID > 0 {
+		for _, s := range m.sites {
+			if s.ID == site.ParentID {
+				breadcrumb = m.st.subtleStyle.Render("  Sites > "+s.Name+" > ") + m.st.titleStyle.Render(site.Name)
+				break
+			}
+		}
+	}
+	if breadcrumb == "" {
+		breadcrumb = m.st.subtleStyle.Render("  Sites > ") + m.st.titleStyle.Render(site.Name)
+	}
+	b.WriteString(breadcrumb + "\n")
+	b.WriteString(m.divider() + "\n")
+
+	row := func(label, value string) {
+		fmt.Fprintf(&b, "  %-16s %s\n", m.st.subtleStyle.Render(label), value)
+	}
+
+	section := func(label string) {
+		b.WriteString("\n" + m.st.subtleStyle.Render("  "+label) + "\n")
+	}
+
+	row("Status", m.fmtStatus(site.Status, site.Paused, m.isMonitorInMaintenance(site.ID)))
+
+	if (site.Status == models.StatusDown || site.Status == models.StatusSSLExp || site.Status == models.StatusLate || site.Status == models.StatusStale) && site.LastError != "" {
+		errWidth := m.termWidth - chromePadH - 19
+		if errWidth < 30 {
+			errWidth = 30
+		}
+		wrapped := lipgloss.NewStyle().Width(errWidth).Render(site.LastError)
+		row("Error", m.st.dangerStyle.Render(wrapped))
+	}
+
+	if site.Type == "http" && site.StatusCode > 0 {
+		row("HTTP Code", strconv.Itoa(site.StatusCode))
+	}
+
+	if (site.Status == models.StatusDown || site.Status == models.StatusSSLExp) && site.LastError != "" {
+		chain := connectionChain(site.LastError, site.Type, site.StatusCode, strings.HasPrefix(site.URL, "https"))
+		if len(chain) > 0 {
+			b.WriteString("\n")
+			for _, step := range chain {
+				var icon string
+				switch step.Status {
+				case stepPassed:
+					icon = m.st.specialStyle.Render("✓")
+				case stepFailed:
+					icon = m.st.dangerStyle.Render("✗")
+				case stepSkipped:
+					icon = m.st.subtleStyle.Render("·")
+				}
+				line := fmt.Sprintf("  %s %-16s", icon, step.Name)
+				if step.Detail != "" {
+					switch step.Status {
+					case stepFailed:
+						line += " " + m.st.dangerStyle.Render(step.Detail)
+					case stepSkipped:
+						line += " " + m.st.subtleStyle.Render(step.Detail)
+					}
+				}
+				b.WriteString(line + "\n")
+			}
+		}
+	}
+
+	if !site.StatusChangedAt.IsZero() {
+		dur := time.Since(site.StatusChangedAt)
+		row("State Since", site.StatusChangedAt.Format("2006-01-02 15:04:05")+" ("+fmtDuration(dur)+")")
+	}
+
+	if !site.LastSuccessAt.IsZero() {
+		ago := time.Since(site.LastSuccessAt)
+		row("Last Success", site.LastSuccessAt.Format("15:04:05")+" ("+fmtDuration(ago)+" ago)")
+	}
+
+	if m.isMonitorInMaintenance(site.ID) {
+		for _, mw := range m.maintenanceWindows {
+			if mw.Type == "maintenance" && (mw.MonitorID == 0 || mw.MonitorID == site.ID || mw.MonitorID == site.ParentID) {
+				row("Maintenance", m.st.maintStyle.Render(mw.Title))
+				break
+			}
+		}
+	}
+
+	section("ENDPOINT")
+	row("Type", site.Type)
+	if site.Type == "push" && site.Token != "" {
+		row("Token", site.Token)
+		row("Push", "curl -X POST -H 'Authorization: Bearer "+site.Token+"' <host>/api/push")
+	}
+	if site.URL != "" {
+		row("URL", site.URL)
+	}
+	if site.Hostname != "" {
+		row("Host", site.Hostname)
+	}
+	if site.Port > 0 {
+		row("Port", strconv.Itoa(site.Port))
+	}
+
+	section("TIMING")
+	row("Interval", fmt.Sprintf("%ds", site.Interval))
+	if site.Timeout > 0 {
+		row("Timeout", fmt.Sprintf("%ds", site.Timeout))
+	}
+	row("Latency", m.fmtLatency(site.Latency))
+	row("Uptime", m.fmtUptime(hist.Statuses))
+	if !site.LastCheck.IsZero() {
+		row("Last Check", m.fmtTimeAgo(site.LastCheck))
+	}
+
+	if site.Type == "http" {
+		section("HTTP")
+		if site.Method != "" && site.Method != "GET" {
+			row("Method", site.Method)
+		}
+		codes := site.AcceptedCodes
+		if codes == "" {
+			codes = "200-299"
+		}
+		row("Codes", codes)
+		row("SSL", m.fmtSSL(site))
+		if site.IgnoreTLS {
+			row("TLS Verify", m.st.dangerStyle.Render("disabled"))
+		}
+	}
+
+	if site.MaxRetries > 0 || site.Regions != "" || site.Description != "" {
+		section("CONFIG")
+		if site.MaxRetries > 0 {
+			row("Retries", m.fmtRetries(site))
+		}
+		if site.Regions != "" {
+			row("Regions", site.Regions)
+		}
+		if site.Description != "" {
+			row("Description", site.Description)
+		}
+	}
+
+	probeResults := m.engine.GetProbeResults(site.ID)
+	if len(probeResults) > 0 {
+		nodeIDs := make([]string, 0, len(probeResults))
+		for id := range probeResults {
+			nodeIDs = append(nodeIDs, id)
+		}
+		sort.Strings(nodeIDs)
+		b.WriteString("\n" + m.st.subtleStyle.Render("  PROBE RESULTS") + "\n")
+		for _, nodeID := range nodeIDs {
+			result := probeResults[nodeID]
+			status := m.st.specialStyle.Render("UP")
+			if !result.IsUp {
+				status = m.st.dangerStyle.Render("DN")
+			}
+			latency := time.Duration(result.LatencyNs).Milliseconds()
+			ago := time.Since(result.CheckedAt).Truncate(time.Second)
+			line := fmt.Sprintf("  %-14s %s  %dms  %s ago", nodeID, status, latency, ago)
+			if !result.IsUp && result.ErrorReason != "" {
+				line += "  " + m.st.dangerStyle.Render(result.ErrorReason)
+			}
+			b.WriteString(line + "\n")
+		}
+	}
+
+	// Loaded on panel-enter (loadDetailCmd) and cached, so View does no DB IO.
+	var stateChanges []models.StateChange
+	if m.detailChangesSiteID == site.ID {
+		stateChanges = m.detailChanges
+	}
+	if len(stateChanges) > 0 {
+		b.WriteString("\n" + m.st.subtleStyle.Render("  STATE CHANGES") + "\n")
+		for i, sc := range stateChanges {
+			ago := fmtDuration(time.Since(sc.ChangedAt))
+			arrow := m.st.subtleStyle.Render(sc.FromStatus) + " → "
+			if sc.ToStatus == string(models.StatusUp) {
+				arrow += m.st.specialStyle.Render(sc.ToStatus)
+			} else {
+				arrow += m.st.dangerStyle.Render(sc.ToStatus)
+			}
+			line := fmt.Sprintf("  %s  %s", arrow, m.st.subtleStyle.Render(ago+" ago"))
+			if dur := computeOutageDuration(stateChanges, i); dur > 0 {
+				line += "  " + m.st.warnStyle.Render("outage "+fmtDuration(dur))
+			}
+			if sc.ErrorReason != "" && sc.ToStatus != string(models.StatusUp) {
+				line += "  " + m.st.dangerStyle.Render(sc.ErrorReason)
+			}
+			b.WriteString(line + "\n")
+		}
+		b.WriteString("  " + m.st.subtleStyle.Render("[h] History") + "\n")
+	}
+
+	b.WriteString(m.divider() + "\n")
+	if site.Type == "push" {
+		b.WriteString("  " + m.zones.Mark("spark-heartbeat", m.heartbeatSparkline(hist.Statuses, detailSparkWidth, "")))
+		if len(hist.Statuses) > 0 {
+			up := 0
+			for _, s := range hist.Statuses {
+				if s {
+					up++
+				}
+			}
+			fmt.Fprintf(&b, "\n  %s %d/%d checks up",
+				m.st.subtleStyle.Render("Heartbeats"),
+				up, len(hist.Statuses))
+		}
+	} else {
+		b.WriteString("  " + m.zones.Mark("spark-latency", m.latencySparkline(hist.Latencies, hist.Statuses, detailSparkWidth, "")))
+		var minL, maxL, total time.Duration
+		count := 0
+		for i, l := range hist.Latencies {
+			if i < len(hist.Statuses) && !hist.Statuses[i] {
+				continue
+			}
+			if count == 0 {
+				minL, maxL = l, l
+			} else if l < minL {
+				minL = l
+			} else if l > maxL {
+				maxL = l
+			}
+			total += l
+			count++
+		}
+		if count > 0 {
+			avg := total / time.Duration(count)
+			fmt.Fprintf(&b, "\n  %s %dms  %s %dms  %s %dms",
+				m.st.subtleStyle.Render("Min"), minL.Milliseconds(),
+				m.st.subtleStyle.Render("Avg"), avg.Milliseconds(),
+				m.st.subtleStyle.Render("Max"), maxL.Milliseconds())
+		}
+	}
+
+	if m.sparkTooltipIdx >= 0 {
+		b.WriteString("\n" + m.renderSparkTooltip(site, hist, detailSparkWidth))
+	}
+
+	b.WriteString("\n")
+	b.WriteString(m.divider() + "\n")
+	b.WriteString(m.st.subtleStyle.Render("  [q/Esc] Back  [e] Edit  [h] History  [s] SLA  [click] Inspect"))
+
+	return lipgloss.NewStyle().Padding(1, 2).Render(b.String())
+}
+
+func (m Model) renderSparkTooltip(site models.Site, hist monitor.SiteHistory, sparkWidth int) string {
+	idx := m.sparkTooltipIdx
+
+	var dataLen int
+	if site.Type == "push" {
+		dataLen = len(hist.Statuses)
+	} else {
+		dataLen = len(hist.Latencies)
+	}
+	if idx < 0 || idx >= dataLen {
+		return ""
+	}
+
+	var parts []string
+
+	checksAgo := dataLen - 1 - idx
+	approxSecs := checksAgo * site.Interval
+	if approxSecs == 0 {
+		parts = append(parts, "latest")
+	} else {
+		parts = append(parts, "~"+fmtDuration(time.Duration(approxSecs)*time.Second)+" ago")
+	}
+
+	if site.Type != "push" && idx < len(hist.Latencies) {
+		parts = append(parts, m.fmtLatency(hist.Latencies[idx]))
+	}
+
+	if idx < len(hist.Statuses) {
+		if hist.Statuses[idx] {
+			parts = append(parts, m.st.specialStyle.Render("UP"))
+		} else {
+			parts = append(parts, m.st.dangerStyle.Render("DOWN"))
+		}
+	}
+
+	sep := m.st.subtleStyle.Render(" | ")
+	pos := m.st.subtleStyle.Render(fmt.Sprintf("[%d/%d]", idx+1, dataLen))
+	return "  " + strings.Join(parts, sep) + "  " + pos
+}

internal/tui/view_detail_inline.go

@@ -1,558 +0,0 @@
-package tui
-
-import (
-	"fmt"
-	"sort"
-	"strings"
-	"time"
-
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/models"
-	"gitea.lerkolabs.com/lerkolabs/uptop/internal/monitor"
-	"github.com/charmbracelet/lipgloss"
-)
-
-func (m Model) viewDetailInline(width, height int) string {
-	if m.cursor >= len(m.sites) {
-		return ""
-	}
-	switch m.detailMode {
-	case detailSLA:
-		return m.viewSLASidebar(width, height)
-	case detailHistory:
-		return m.viewHistorySidebar(width, height)
-	default:
-		site := m.sites[m.cursor]
-		hist, _ := m.engine.GetHistory(site.ID)
-		return m.buildDetailContent(site, hist, width, false)
-	}
-}
-
-func (m Model) viewDetailFullscreen() string {
-	if m.cursor >= len(m.sites) {
-		return ""
-	}
-
-	availW := m.termWidth - chromePadH
-	site := m.sites[m.cursor]
-
-	var title string
-	switch m.detailMode {
-	case detailSLA:
-		title = "SLA · " + site.Name
-	case detailHistory:
-		title = "History · " + site.Name
-	default:
-		title = site.Name
-	}
-
-	if site.ParentID > 0 {
-		for _, s := range m.sites {
-			if s.ID == site.ParentID {
-				title = s.Name + " > " + title
-				break
-			}
-		}
-	}
-
-	innerW := availW - 2
-
-	var content string
-	switch m.detailMode {
-	case detailSLA:
-		content = m.viewSLASidebar(innerW, 0)
-	case detailHistory:
-		content = m.viewHistorySidebar(innerW, 0)
-	default:
-		hist, _ := m.engine.GetHistory(site.ID)
-		content = m.buildDetailContent(site, hist, innerW, true)
-	}
-
-	footer := m.detailFooter(innerW)
-
-	panelH := m.termHeight - chromePadV
-	if panelH < 10 {
-		panelH = 10
-	}
-
-	return lipgloss.NewStyle().Padding(1, 2).Render(
-		m.titledPanelH(title, content, footer, availW, panelH, m.detailScrollOffset, scrollbar{}, true))
-}
-
-func (m Model) buildDetailContent(site models.Site, hist monitor.SiteHistory, width int, fullscreen bool) string {
-	dot := m.st.subtleStyle.Render(" · ")
-	label := m.st.subtleStyle
-	innerW := width - 4
-	if innerW < 20 {
-		innerW = 20
-	}
-
-	var b strings.Builder
-
-	// Status + latency + last check + state since
-	status := m.fmtStatus(site.Status, site.Paused, m.isMonitorInMaintenance(site.ID))
-	statusParts := []string{status}
-	if site.Latency > 0 {
-		statusParts = append(statusParts, m.fmtLatency(site.Latency))
-	}
-	if !site.LastCheck.IsZero() {
-		statusParts = append(statusParts, m.fmtTimeAgo(site.LastCheck))
-	}
-	if !site.StatusChangedAt.IsZero() {
-		dur := time.Since(site.StatusChangedAt)
-		statusParts = append(statusParts, label.Render("for")+" "+fmtDuration(dur))
-	}
-	b.WriteString("  " + strings.Join(statusParts, dot) + "\n")
-
-	// Type-specific details
-	typeParts := m.detailTypeLine(site)
-	if len(typeParts) > 0 {
-		b.WriteString("  " + strings.Join(typeParts, dot) + "\n")
-	}
-
-	// Extended endpoint fields
-	m.writeEndpointFields(&b, site, label, innerW, fullscreen)
-
-	// Uptime + retries + last success
-	uptimeStr := m.fmtUptime(hist.Statuses)
-	if m.isMonitorInMaintenance(site.ID) {
-		uptimeStr = m.st.subtleStyle.Render("—")
-	}
-	uptimeParts := []string{label.Render("Uptime") + " " + uptimeStr}
-	if site.Type != "group" && site.MaxRetries > 0 {
-		uptimeParts = append(uptimeParts, label.Render("Retries")+" "+m.fmtRetries(site))
-	}
-	if site.Type != "push" && !site.LastSuccessAt.IsZero() {
-		uptimeParts = append(uptimeParts, label.Render("Last OK")+" "+m.fmtTimeAgo(site.LastSuccessAt))
-	}
-	b.WriteString("  " + strings.Join(uptimeParts, dot) + "\n")
-
-	// Maintenance window name
-	if m.isMonitorInMaintenance(site.ID) {
-		for _, mw := range m.maintenanceWindows {
-			if mw.Type == "maintenance" && (mw.MonitorID == 0 || mw.MonitorID == site.ID || mw.MonitorID == site.ParentID) {
-				b.WriteString("  " + label.Render("Maint") + " " + m.st.maintStyle.Render(mw.Title) + "\n")
-				break
-			}
-		}
-	}
-
-	// Error line
-	if (site.Status == models.StatusDown || site.Status == models.StatusSSLExp ||
-		site.Status == models.StatusLate || site.Status == models.StatusStale) && site.LastError != "" {
-		errW := innerW
-		if errW < 20 {
-			errW = 20
-		}
-		b.WriteString("  " + label.Render("Error") + " " + m.st.dangerStyle.Render(limitStr(site.LastError, errW)) + "\n")
-	}
-
-	// Connection chain
-	if (site.Status == models.StatusDown || site.Status == models.StatusSSLExp) && site.LastError != "" {
-		chain := connectionChain(site.LastError, site.Type, site.StatusCode, strings.HasPrefix(site.URL, "https"))
-		if len(chain) > 0 {
-			b.WriteString("\n")
-			for _, step := range chain {
-				var icon string
-				switch step.Status {
-				case stepPassed:
-					icon = m.st.specialStyle.Render("✓")
-				case stepFailed:
-					icon = m.st.dangerStyle.Render("✗")
-				case stepSkipped:
-					icon = m.st.subtleStyle.Render("·")
-				}
-				line := fmt.Sprintf("  %s %-16s", icon, step.Name)
-				if step.Detail != "" {
-					switch step.Status {
-					case stepFailed:
-						line += " " + m.st.dangerStyle.Render(step.Detail)
-					case stepSkipped:
-						line += " " + m.st.subtleStyle.Render(step.Detail)
-					}
-				}
-				b.WriteString(line + "\n")
-			}
-		}
-	}
-
-	b.WriteString("\n")
-
-	// Probe results
-	probeResults := m.engine.GetProbeResults(site.ID)
-	if len(probeResults) > 0 {
-		nodeIDs := make([]string, 0, len(probeResults))
-		for id := range probeResults {
-			nodeIDs = append(nodeIDs, id)
-		}
-		sort.Strings(nodeIDs)
-		for _, nodeID := range nodeIDs {
-			result := probeResults[nodeID]
-			probeStatus := m.st.specialStyle.Render("UP")
-			if !result.IsUp {
-				probeStatus = m.st.dangerStyle.Render("DN")
-			}
-			latency := time.Duration(result.LatencyNs).Milliseconds()
-			ago := time.Since(result.CheckedAt).Truncate(time.Second)
-			line := fmt.Sprintf("  %-14s %s  %dms  %s ago", nodeID, probeStatus, latency, ago)
-			if !result.IsUp && result.ErrorReason != "" {
-				line += "  " + m.st.dangerStyle.Render(result.ErrorReason)
-			}
-			b.WriteString(line + "\n")
-		}
-		b.WriteString("\n")
-	}
-
-	// Latency chart
-	if len(hist.Latencies) > 0 {
-		chart := m.latencyChart(hist.Latencies, hist.Statuses, innerW, 3)
-		if chart != "" {
-			b.WriteString(chart + "\n")
-		}
-	}
-
-	// 30d uptime timeline
-	if len(m.detailDailyDays) > 0 && m.detailChangesSiteID == site.ID {
-		b.WriteString("  " + label.Render("30d") + " " + m.uptimeTimeline(m.detailDailyDays, innerW) + "\n")
-	}
-
-	// Sparkline + min/avg/max
-	if site.Type != "push" && len(hist.Latencies) > 0 {
-		b.WriteString("  " + m.latencySparkline(hist.Latencies, hist.Statuses, innerW, nil) + "\n")
-		var minL, maxL, total time.Duration
-		count := 0
-		for i, l := range hist.Latencies {
-			if i < len(hist.Statuses) && !hist.Statuses[i] {
-				continue
-			}
-			if count == 0 {
-				minL, maxL = l, l
-			} else if l < minL {
-				minL = l
-			} else if l > maxL {
-				maxL = l
-			}
-			total += l
-			count++
-		}
-		if count > 0 {
-			avg := total / time.Duration(count)
-			fmt.Fprintf(&b, "  %s %dms  %s %dms  %s %dms\n",
-				label.Render("Min"), minL.Milliseconds(),
-				label.Render("Avg"), avg.Milliseconds(),
-				label.Render("Max"), maxL.Milliseconds())
-		}
-	}
-
-	// Latency histogram
-	if site.Type != "push" && len(hist.Latencies) > 5 {
-		histContent := m.latencyHistogram(hist.Latencies, hist.Statuses, innerW)
-		if histContent != "" {
-			b.WriteString("\n")
-			b.WriteString(histContent)
-		}
-	}
-
-	b.WriteString("\n")
-
-	// State changes
-	var stateChanges []models.StateChange
-	if m.detailChangesSiteID == site.ID {
-		stateChanges = m.detailChanges
-	}
-	if len(stateChanges) > 0 {
-		limit := 5
-		if len(stateChanges) < limit {
-			limit = len(stateChanges)
-		}
-		for _, sc := range stateChanges[:limit] {
-			ago := fmtDuration(time.Since(sc.ChangedAt))
-			arrow := m.st.subtleStyle.Render("→")
-			from := m.fmtStatusWord(sc.FromStatus)
-			to := m.fmtStatusWord(sc.ToStatus)
-			entry := from + " " + arrow + " " + to + " " + label.Render(ago+" ago")
-			if sc.ErrorReason != "" {
-				reasonW := innerW - 25
-				if reasonW < 15 {
-					reasonW = 15
-				}
-				entry += "  " + m.st.dangerStyle.Render(limitStr(sc.ErrorReason, reasonW))
-			}
-			b.WriteString("  " + entry + "\n")
-		}
-	} else {
-		b.WriteString("  " + label.Render("No state changes") + "\n")
-	}
-
-	return lipgloss.NewStyle().Width(width).MaxWidth(width).Render(b.String())
-}
-
-func (m Model) writeEndpointFields(b *strings.Builder, site models.Site, label lipgloss.Style, innerW int, fullscreen bool) {
-	dot := m.st.subtleStyle.Render(" · ")
-	var fields []string
-
-	if site.Interval > 0 {
-		fields = append(fields, label.Render("Every")+" "+fmt.Sprintf("%ds", site.Interval))
-	}
-	if site.Timeout > 0 {
-		fields = append(fields, label.Render("Timeout")+" "+fmt.Sprintf("%ds", site.Timeout))
-	}
-	if site.Type == "http" && site.Method != "" && site.Method != "GET" {
-		fields = append(fields, label.Render("Method")+" "+site.Method)
-	}
-	if site.Type == "http" {
-		codes := site.AcceptedCodes
-		if codes == "" {
-			codes = "200-299"
-		}
-		fields = append(fields, label.Render("Codes")+" "+codes)
-	}
-	if site.Regions != "" {
-		fields = append(fields, label.Render("Regions")+" "+site.Regions)
-	}
-
-	if len(fields) > 0 {
-		b.WriteString("  " + strings.Join(fields, dot) + "\n")
-	}
-
-	if site.Description != "" {
-		maxDescW := innerW
-		if !fullscreen && maxDescW > 60 {
-			maxDescW = 60
-		}
-		b.WriteString("  " + label.Render(limitStr(site.Description, maxDescW)) + "\n")
-	}
-
-	if site.Type == "push" && site.Token != "" {
-		b.WriteString("  " + label.Render("Token") + " " + site.Token + "\n")
-	}
-}
-
-func (m Model) detailTypeLine(site models.Site) []string {
-	label := m.st.subtleStyle
-	var parts []string
-
-	switch site.Type {
-	case "http":
-		if site.StatusCode > 0 {
-			codeStr := fmt.Sprintf("HTTP %d", site.StatusCode)
-			if site.StatusCode >= httpErrorThreshold {
-				parts = append(parts, m.st.dangerStyle.Render(codeStr))
-			} else {
-				parts = append(parts, m.st.specialStyle.Render(codeStr))
-			}
-		}
-		if site.CheckSSL && site.HasSSL {
-			days := int(time.Until(site.CertExpiry).Hours() / 24)
-			sslStr := fmt.Sprintf("SSL %dd", days)
-			switch {
-			case days <= 0:
-				parts = append(parts, m.st.dangerStyle.Render("SSL EXPIRED"))
-			case days <= site.ExpiryThreshold:
-				parts = append(parts, m.st.warnStyle.Render(sslStr))
-			default:
-				parts = append(parts, m.st.specialStyle.Render(sslStr))
-			}
-		}
-		if site.URL != "" {
-			parts = append(parts, label.Render(limitStr(site.URL, 40)))
-		}
-	case "push":
-		parts = append(parts, label.Render("Push"))
-		if site.Interval > 0 {
-			parts = append(parts, label.Render(fmt.Sprintf("every %s", fmtDuration(time.Duration(site.Interval)*time.Second))))
-		}
-		if !site.LastSuccessAt.IsZero() {
-			parts = append(parts, label.Render("last")+" "+m.fmtTimeAgo(site.LastSuccessAt))
-		}
-	case "ping":
-		parts = append(parts, label.Render("Ping"))
-		if site.Hostname != "" {
-			parts = append(parts, label.Render(site.Hostname))
-		}
-	case "port":
-		parts = append(parts, label.Render("Port"))
-		if site.Hostname != "" {
-			target := site.Hostname
-			if site.Port > 0 {
-				target = fmt.Sprintf("%s:%d", site.Hostname, site.Port)
-			}
-			parts = append(parts, label.Render(target))
-		}
-	case "dns":
-		parts = append(parts, label.Render("DNS"))
-		if site.DNSResolveType != "" {
-			parts = append(parts, label.Render(site.DNSResolveType))
-		}
-		if site.DNSServer != "" {
-			parts = append(parts, label.Render(site.DNSServer))
-		}
-	}
-
-	return parts
-}
-
-func (m Model) detailFooter(width int) string {
-	dot := m.st.subtleStyle.Render(" · ")
-	var parts []string
-
-	switch m.detailMode {
-	case detailSLA:
-		for i, p := range slaPeriods {
-			if i == m.slaPeriodIdx {
-				parts = append(parts, m.st.titleStyle.Render(p.key)+" "+m.st.titleStyle.Render(p.label))
-			} else {
-				parts = append(parts, m.hotkey(p.key, p.label))
-			}
-		}
-		parts = append(parts, m.hotkey("Esc", "Back"))
-	case detailHistory:
-		parts = append(parts, m.hotkey("Esc", "Back"))
-	default:
-		parts = append(parts, m.hotkey("e", "Edit"), m.hotkey("h", "History"), m.hotkey("s", "SLA"), m.hotkey("Esc", "Back"))
-	}
-
-	content := "  " + strings.Join(parts, dot)
-	return lipgloss.NewStyle().Width(width).MaxWidth(width).Render(content)
-}
-
-func (m Model) fmtStatusWord(status string) string {
-	switch status {
-	case "DOWN", "SSL EXP":
-		return m.st.dangerStyle.Render(status)
-	case "UP":
-		return m.st.specialStyle.Render("UP")
-	case "LATE":
-		return m.st.warnStyle.Render("LATE")
-	case "STALE":
-		return m.st.staleStyle.Render("STALE")
-	case "PENDING":
-		return m.st.subtleStyle.Render("PENDING")
-	case "PAUSED":
-		return m.st.warnStyle.Render("PAUSED")
-	default:
-		return m.st.subtleStyle.Render(status)
-	}
-}
-
-func (m Model) viewSLASidebar(width, _ int) string {
-	var b strings.Builder
-	label := m.st.subtleStyle
-	innerW := width - 4
-	if innerW < 20 {
-		innerW = 20
-	}
-
-	period := slaPeriods[m.slaPeriodIdx]
-	b.WriteString("  " + label.Render("Period: Last "+period.label) + "\n\n")
-
-	r := m.slaReport
-	barWidth := innerW - 25
-	if barWidth < 10 {
-		barWidth = 10
-	}
-	bar := m.uptimeBar(r.UptimePct, barWidth)
-	uptimeColor := m.st.specialStyle
-	if r.UptimePct < uptimeExcellentPct {
-		uptimeColor = m.st.warnStyle
-	}
-	if r.UptimePct < uptimeGoodPct {
-		uptimeColor = m.st.dangerStyle
-	}
-	fmt.Fprintf(&b, "  %-14s %s  %s\n", label.Render("Uptime"), uptimeColor.Render(fmtPct(r.UptimePct)+"%"), bar)
-	fmt.Fprintf(&b, "  %-14s %s\n", label.Render("Downtime"), fmtDuration(r.Downtime))
-	fmt.Fprintf(&b, "  %-14s %d\n", label.Render("Outages"), r.OutageCount)
-
-	if r.OutageCount > 0 {
-		fmt.Fprintf(&b, "  %-14s %s\n", label.Render("Longest"), fmtDuration(r.LongestOut))
-		fmt.Fprintf(&b, "  %-14s %s\n", label.Render("MTTR"), fmtDuration(r.MTTR))
-		fmt.Fprintf(&b, "  %-14s %s\n", label.Render("MTBF"), fmtDuration(r.MTBF))
-	}
-
-	b.WriteString("\n")
-
-	if len(m.slaDailyBreakdown) > 0 {
-		b.WriteString("  " + m.st.titleStyle.Render("DAILY BREAKDOWN") + "\n")
-		dayBarW := innerW - 20
-		if dayBarW < 10 {
-			dayBarW = 10
-		}
-		for _, day := range m.slaDailyBreakdown {
-			dateStr := day.Date.Format("Jan 02")
-			dayBar := m.uptimeBar(day.UptimePct, dayBarW)
-			pctStr := fmtPct(day.UptimePct) + "%"
-			color := m.st.specialStyle
-			if day.UptimePct < uptimeExcellentPct {
-				color = m.st.warnStyle
-			}
-			if day.UptimePct < uptimeGoodPct {
-				color = m.st.dangerStyle
-			}
-			fmt.Fprintf(&b, "  %-8s %s %s\n", label.Render(dateStr), dayBar, color.Render(pctStr))
-		}
-	}
-
-	return lipgloss.NewStyle().Width(width).MaxWidth(width).Render(b.String())
-}
-
-func (m Model) viewHistorySidebar(width, _ int) string {
-	var b strings.Builder
-	label := m.st.subtleStyle
-	innerW := width - 4
-	if innerW < 20 {
-		innerW = 20
-	}
-
-	sparkline := m.stateChangeSparkline(m.historyChanges, innerW)
-	if sparkline != "" {
-		b.WriteString("  " + sparkline + "\n\n")
-	}
-
-	if len(m.historyChanges) == 0 {
-		b.WriteString("  " + label.Render("No state changes recorded") + "\n")
-	} else {
-		reasonW := innerW - 45
-		if reasonW < 10 {
-			reasonW = 10
-		}
-		for i, sc := range m.historyChanges {
-			ts := sc.ChangedAt.Format("01/02 15:04")
-
-			arrow := label.Render(sc.FromStatus) + " → "
-			switch sc.ToStatus {
-			case string(models.StatusUp):
-				arrow += m.st.specialStyle.Render(sc.ToStatus)
-			case string(models.StatusLate):
-				arrow += m.st.warnStyle.Render(sc.ToStatus)
-			case string(models.StatusStale):
-				arrow += m.st.staleStyle.Render(sc.ToStatus)
-			default:
-				arrow += m.st.dangerStyle.Render(sc.ToStatus)
-			}
-
-			durStr := ""
-			if dur := computeOutageDuration(m.historyChanges, i); dur > 0 {
-				durStr = m.st.warnStyle.Render(fmtDuration(dur))
-			}
-
-			reason := ""
-			if sc.ErrorReason != "" && sc.ToStatus != string(models.StatusUp) {
-				reason = m.st.dangerStyle.Render(limitStr(sc.ErrorReason, reasonW))
-			}
-
-			fmt.Fprintf(&b, "  %-12s %s  %s %s\n", ts, arrow, durStr, reason)
-		}
-	}
-
-	b.WriteString("\n")
-
-	stats := computeHistoryStats(m.historyChanges)
-	statParts := []string{fmt.Sprintf("%d events", stats.totalEvents)}
-	if stats.outageCount > 0 {
-		statParts = append(statParts, fmt.Sprintf("%d outages", stats.outageCount))
-		avg := stats.totalDowntime / time.Duration(stats.outageCount)
-		statParts = append(statParts, "avg "+fmtDuration(avg))
-	}
-	b.WriteString("  " + label.Render(strings.Join(statParts, " │ ")) + "\n")
-
-	return lipgloss.NewStyle().Width(width).MaxWidth(width).Render(b.String())
-}