Skip to main content

Production Checklist

Use this checklist before and after deploying AgentHub in a shared environment.

Preflight Checklist​

  • Define explicit safe_paths for each team/project root.
  • Validate worktree.default_root exists and is writable.
  • Confirm runtime user permissions for repository paths.
  • Confirm Node/Rust build artifacts are reproducible in CI.
  • Confirm rollback plan (previous binary + config snapshot).

Security Baseline​

  • Run AgentHub as a non-root user.
  • Keep network exposure minimal (private network + reverse proxy).
  • Use TLS at the edge.
  • Enforce strong user credentials and periodic rotation.
  • Avoid putting sensitive directories under safe_paths.

Data and Backup​

AgentHub persists runtime data under ~/.agenthub/ by default.

Minimum backup targets:

  • ~/.agenthub/agenthub.db
  • deployment config.toml

Recommended cadence:

  • Daily snapshot for active environments
  • Extra snapshot before each upgrade

Upgrade Runbook​

  1. Stop new task submissions.
  2. Backup agenthub.db and config.toml.
  3. Deploy new binary/config.
  4. Restart service.
  5. Run smoke checks:
    • login
    • create/start one test agent
    • session replay after refresh
  6. Monitor logs for startup/runtime errors.

Failure Rollback Rule​

If smoke checks fail or critical flows regress:

  1. Restore previous binary.
  2. Restore previous config.
  3. Restore database snapshot when schema compatibility is uncertain.
  4. Re-run smoke checks before reopening service.