Skip to main content

Husky dashboard guide

Husky includes an embedded dashboard built from the web/ app and served by the daemon. No separate frontend service is required in production.

Accessing the dashboard​

Start the daemon, then run:

husky dash

The CLI opens the HTTP address recorded in <data>/api.addr.

What the dashboard provides today​

Top bar / daemon controls​

The dashboard shows daemon connectivity, uptime, and version information.

Current daemon actions include:

  • reload config
  • stop daemon gracefully or forcefully
  • inspect daemon info such as PID, config path, DB path, and counts

Jobs tab​

The Jobs tab is the main operator surface.

It provides:

  • job list with description, frequency, timezone, tags, and schedule state
  • current running / paused state
  • tag filter dropdown
  • tag health summary strip
  • per-job run, cancel, pause, and resume actions
  • expandable recent-run history per job
  • run pills that open detailed run views

Run detail view​

The dashboard has a dedicated run detail page that shows:

  • status, attempt, trigger, triggered-by, and reason
  • start / finish / duration
  • exit code
  • SLA state
  • healthcheck status
  • full log viewer
  • captured output variables
  • links to previous and next runs for the same job

Log viewing​

The log viewer supports:

  • historical log retrieval for completed runs
  • live WebSocket streaming for running jobs
  • optional healthcheck stream visibility
  • stream-aware rendering for stdout, stderr, and healthcheck lines

Audit tab​

The Audit view exposes searchable run history across jobs.

Filters include:

  • job
  • status
  • trigger
  • tag
  • date / since filter
  • reason text

Rows can open a run detail workflow so operators can move from audit metadata to logs and outputs quickly.

Outputs / Data views​

The dashboard includes database-backed views for:

  • run_outputs
  • job_runs
  • run_logs
  • job_state
  • alerts

These are especially useful for pipeline debugging and notification troubleshooting.

DAG tab​

The DAG view renders job dependencies and supports:

  • graph navigation
  • opening job-related run views
  • cycle/correlation-oriented workflows when following a pipeline execution

Health and integrations​

The dashboard exposes:

  • SLA and health-oriented summaries
  • integration status and test actions
  • config viewing and save/reload workflows

API relationship​

The dashboard is a thin client over the Husky HTTP API.

Important endpoint groups include:

  • /api/jobs
  • /api/runs/:id
  • /api/audit
  • /api/tags
  • /api/dag
  • /api/db/*
  • /api/config*
  • /api/daemon/*
  • /ws/logs/:run_id

See configuration.md and operations.md for surrounding runtime details.

Dashboard deployment model​

  • frontend source lives in web/
  • built assets are generated into internal/api/dashboard/
  • Go embeds those assets into the husky binary
  • the daemon serves the embedded bundle directly

That means:

  • production does not need Node.js
  • there is no separate dashboard package to deploy
  • dashboard version always matches the daemon binary version

Dashboard-focused troubleshooting​

If husky dash does not work:

  1. verify the daemon is running with husky status
  2. inspect <data>/api.addr
  3. check whether huskyd.yaml bound the API to a different address
  4. if auth is enabled, confirm the browser or client has valid credentials
  5. review daemon logs for API startup failures

If live logs do not stream:

  • confirm the run is still active
  • confirm WebSocket access is allowed under your auth and RBAC config
  • check whether a reverse proxy is stripping WebSocket upgrades

If the dashboard loads but data is empty:

  • confirm the daemon was started with the expected husky.yaml
  • confirm huskyd.yaml base path or auth settings are consistent with how you access the dashboard