User documentation¶
End-user docs for Kneo Agent Dashboard — for operators, SREs, and reviewers running deployed Kneo agents.
For the top-level docs index see docs/README.md.
Reading order¶
Daily operation¶
- Quickstart — run the BFF + SPA against a Kneo Agent Platform, ~10 min. (Prefer a guided, timed end-to-end path? The tutorial walks deploy → connect → launch → live trace → HITL → audit in under an hour — 0.8.0.)
- Connecting — the
kneo-clientprofile, environments, the state store. - Runs & debugging — Overview, the runs list, run detail (trace tail · checkpoints · replay/diff · graph · chain), run control, sessions, compare, launch.
- Human-in-the-loop — the reviewer queue, richer decisions, idempotent + bulk resume, near-deadline alerts.
- Audit & health — audit search/filter; health probes + subsystems.
- Policies & credentials — environment policy (view / dry-run / apply) and the presence-only credential inventory.
- Cost & spend — the approximate blended per-environment spend figure and the price book.
- Settings — the four dashboard-own cards (Connections · Access · Pricing · Preferences).
Operating in production¶
- Deployment — run the BFF (container /
pip install), state store (SQLite vs Postgres), liveness/readiness probes, non-root. - Observability & monitoring — the
/metricssurface, what to watch, alerting. - Security hardening & auth operations — the pre-launch checklist, OIDC vs static mode, sessions + secret rotation, reverse-proxy/TLS, and what an expired-session / cross-origin / role-escalation attempt does.
- Backup, restore & recovery — backing up the state store, the
fail-safe restore procedure (
recover→--reconcile), rollback, RPO/RTO, DR checklist. - Data handling & retention — what the local store holds, retention, deletion limits.
- Upgrade — moving between versions (BFF + SPA + state-store migrations).
- Troubleshooting — common failures and how to trace them by request id.
Reference¶
- Environment variables — every setting the BFF reads, grouped by concern.
- CLI reference — the
kneo-dashcommands (serve,recover,prune). - API reference — the browsable
/apicontract (generated fromopenapi.json). - Examples & coverage matrix — worked scenarios + the page/router/capability→doc matrix.
- Accessibility — keyboard navigation, screen-reader support, contrast.
Covers the 0.7.0 operator surface — the full 0.4.0–0.6.0 platform surface (OIDC auth, server-side RBAC, the live env switcher, approximate cost, Settings, the recovery / break-glass path, the
/metricstelemetry surface, and server-truth capability gating) plus 0.7.0's documentation-excellence + accessibility work: the new cost / settings / observability / data-handling / CLI / upgrade / troubleshooting / accessibility / examples guides, the browsable API reference, current-UI screenshots, and the WCAG-AA contrast pass. (Ships:: 3 - Alpha.)
Pages¶
- Quickstart
- Tutorial — zero to operating, timed (0.8.0)
- Connecting to the platform
- Runs & debugging
- Human-in-the-loop
- Audit & health
- Observability & monitoring
- Cost & spend
- Policies & credentials
- Settings
- Deployment guide
- CLI reference
- Environment-variable reference
- Security hardening & auth operations
- Accessibility
- Data handling & retention
- Backup, restore & recovery
- Upgrading
- Troubleshooting
- Examples & coverage
- API reference (BFF
/api/*)