Skip to content

STATUS.md

Latest commit

 

History

History
89 lines (77 loc) · 7.42 KB

STATUS.md

File metadata and controls

89 lines (77 loc) · 7.42 KB

STATUS.md — Private Dashboard

Current State

Phase: Deployed to staging, collector live, data retention active
Tests: 202 Rust + 187 Python SDK integration
**Last Updated: 2026-02-18 14:55 UTC

What's Done

  • ✅ Project scaffold (Rust/Rocket + React/Tailwind)

  • ✅ SQLite database with stats table + config table

  • ✅ POST /api/v1/stats — batch submit with auth (max 100, key validation)

  • ✅ GET /api/v1/stats — all metrics with trend data (24h/7d/30d/90d)

  • ✅ GET /api/v1/stats/:key — single stat history with period filter

  • ✅ GET /api/v1/health — health check with stats count, retention info, oldest stat

  • ✅ POST /api/v1/stats/prune — manual data retention trigger (auth required)

  • ✅ Auto-prune on startup (90-day retention, deletes old stats automatically)

  • ✅ GET /llms.txt — AI-readable API summary (updated with prune docs)

  • ✅ GET /openapi.json — full OpenAPI 3.0.3 spec (updated with prune endpoint)

  • ✅ Auto-generated manage key on first run

  • ✅ Frontend: dark theme dashboard with stat cards, sparklines, trend badges

  • ✅ Frontend: responsive grid, auto-refresh 60s, empty state

  • ✅ 94 tests (auth, submit, query, validation, trends, batch limits, key labels, llms.txt, openapi, edge cases, prune, skills discovery, sparklines, alerts)

  • ✅ Dockerfile (multi-stage: frontend + backend)

  • ✅ docker-compose.yml (port 3008)

  • ✅ Deployed to staging (192.168.0.79:3008)

  • ✅ GitHub repo created, pushed to main

  • ✅ DESIGN.md with full API spec

  • ✅ DB backup script already includes private-dashboard

  • ✅ Collector cron (every 30 min) — gathers 10 metrics from workspace state files

  • ✅ Collector script: scripts/dashboard-collector.py (in workspace)

  • ✅ Metric labels for all collector keys (siblings_active, moltbook_health, moltbook_my_posts, twitter_accounts)

  • ✅ Edge case tests: negative/zero/large/fractional values, special chars, invalid JSON, missing fields, large metadata, all periods, rapid writes, 50-key batch

  • ✅ Data retention: 90-day auto-prune on startup + manual prune endpoint + 9 prune tests

  • ✅ UI/UX overhaul: viewport-filling layout, responsive sparklines, improved labels, better trend display

  • ✅ Custom SVG logo and favicon (ascending bar chart design, dark theme matching)

  • ✅ Rename: "HNR Dashboard" → "The Pack" (Agent Operations) — represents Nanook + siblings

  • ✅ Per-period sparklines: 7d/30d/90d fetch actual history data from API

  • ✅ Sparkline hover: crosshair + value tooltip on mouse hover

  • ✅ Relative time labels: "5m ago" instead of raw timestamps

  • ✅ Backend trend fix: falls back to earliest point in window (fixes null trends for new data)

  • ✅ Better null trend display: "no data yet" instead of "collecting…"

  • ✅ Responsive mobile layout: 1-col mobile, 2-col tablet, viewport-filling desktop

  • ✅ GitHub Actions CI/CD — cargo test + Docker build + push to ghcr.io (a425e2a)

  • ✅ Touch-friendly period buttons on mobile

  • ✅ Stacked header on small screens, responsive font sizes

  • ✅ Responsive fix (ee54118): flex-1/min-h-0 → lg:flex-1/lg:min-h-0 on stats wrapper, groups, and grids. Mobile/tablet now size naturally and scroll; desktop unchanged.

  • ✅ Unit suffixes — contextual units next to metric values (agents, commits, tests, repos, etc.) (aafae8a)

  • ✅ Already on /mylinks page (hnrstage.xyz/mylinks) — first card with all links

  • ✅ Metric grouping — Development, Network, Moltbook, Social sections with labeled dividers (7403127)

  • ✅ Binary metric display — moltbook_health shows "Healthy"/"Down" with color instead of 1/0 (7403127)

  • ✅ Trend alerts — pulsing dot + glow border for significant 24h changes (±10% alert, ±25% hot with ⚡) (f0ea09d)

  • ✅ Metric detail view — click any card for full modal with interactive chart, trend summaries, data table (6e14145)

  • ✅ Custom date range — "Custom" period option with date pickers; backend accepts start/end params (ISO-8601 or YYYY-MM-DD); 6 new tests (74f445a)

  • ✅ DELETE /api/v1/stats/:key — Auth-required metric cleanup endpoint; deletes all data for a key; 404 for nonexistent, 403 for bad auth; 6 new tests (7e5376f)

  • ✅ Period button click fix — stopPropagation on StatCard period buttons so clicking period doesn't open modal (c9436ab)

  • ✅ CSV export — Download button in MetricDetail data table exports timestamp+value pairs as CSV (c9436ab)

  • ✅ Alert history log — alert_log table auto-records significant metric changes on submit; GET /api/v1/alerts with key filter + limit; 10%=alert, 25%=hot, 6h debounce; 9 new tests (2465027)

  • ✅ Frontend alert history panel — 🔔 Recent Alerts section below stats grid; color-coded by level (hot=⚡, alert=dot) and direction; expand/collapse for >5 alerts; responsive; auto-refreshes with 60s interval (fbeb7db)

  • ✅ cron_jobs_active metric — label, unit suffix ("jobs"), Network group in frontend (17cc44e)

  • ✅ Kanban board metrics — 6 new metrics (backlog, up_next, in_progress, review, done, active) with "Work Queue" group, labels, units, tests (1ce57c8)

  • ✅ Full-viewport modal on mobile — MetricDetail modal now fills entire screen on mobile (no padding, no rounded corners, full height); desktop unchanged (4a2a60e)

  • ✅ Expanded test coverage — 24 new tests: stats alphabetical ordering, full response field validation (stats + health), sparkline downsampling/passthrough, trend edge cases (zero start, no prior data), key_label fallback + all known keys, seq monotonicity, custom date range partial params, OpenAPI spec structure, prune/alert independence, alert label mapping, limit clamping, mixed batch validation, multiple submits latest-value, history ordering, delete cleanup, auth method coverage, health keys/stats distinction, llms.txt endpoint docs. 92 total tests (9a6bf92).

  • ✅ Python SDK — Complete zero-dependency Python client library (sdk/python/dashboard.py). Covers all API endpoints (health, stats, history, submit, delete, prune, alerts, discovery). Dict shorthand for ergonomic submission (dash.submit({"tests_total": 1500})). Typed error hierarchy (AuthError, NotFoundError, ValidationError, RateLimitError). Convenience helpers (get_value, get_trend, keys, is_healthy, hot_alerts). Env config (DASHBOARD_URL, DASHBOARD_KEY). 102 integration tests. SDK README with full docs and examples.

What's Next

  1. Staging domain — Add dashboard. to Cloudflare wildcard (Jordan action)

⚠️ Gotchas

  • CI workflow push confirmed working (Feb 13)
  • Manage key is auto-generated on first run and printed to stdout — save it
  • Current manage key on staging: dash_0e54dee0985e417b8ecb78b3607ad816
  • Frontend requires bun for Docker build (same pattern as other HNR projects)
  • openapi.json must be COPY'd in Dockerfile backend stage (include_str! needs it at compile time)
  • Staging runs ghcr.io image + Watchtower label enabled (auto-pulls :dev every 5 min)
  • Collector cron ID: 31f1ab2e-c191-4872-a9ce-f746e5d74928
  • moltbook_health reads 0 when platformHealth.status != "healthy" (check moltbook-state.json)
  • Data retention: 90 days default, runs on startup, manual trigger via POST /api/v1/stats/prune

Incoming Directions (Work Queue)

  • 2026-02-14 01:10 UTC (Jordan): Responsive layout is not working properly. Please re-investigate. (task 98f0acf7) → Fixed in ee54118: root cause was unconditional flex-1/min-h-0 crushing groups on mobile.