diff --git a/docs/user-guide/img/agent-hitl-approve.png b/docs/user-guide/img/agent-hitl-approve.png new file mode 100644 index 00000000..d0d4bd2a Binary files /dev/null and b/docs/user-guide/img/agent-hitl-approve.png differ diff --git a/docs/user-guide/img/inspect-artifacts-panel.png b/docs/user-guide/img/inspect-artifacts-panel.png new file mode 100644 index 00000000..f8ed83f9 Binary files /dev/null and b/docs/user-guide/img/inspect-artifacts-panel.png differ diff --git a/docs/user-guide/img/kpi-strip.png b/docs/user-guide/img/kpi-strip.png new file mode 100644 index 00000000..abb73afd Binary files /dev/null and b/docs/user-guide/img/kpi-strip.png differ diff --git a/docs/user-guide/img/ops-snapshot-tiles.png b/docs/user-guide/img/ops-snapshot-tiles.png new file mode 100644 index 00000000..35c1724d Binary files /dev/null and b/docs/user-guide/img/ops-snapshot-tiles.png differ diff --git a/docs/user-guide/img/run-history-strip.png b/docs/user-guide/img/run-history-strip.png new file mode 100644 index 00000000..d926a428 Binary files /dev/null and b/docs/user-guide/img/run-history-strip.png differ diff --git a/docs/user-guide/showcase-walkthrough.md b/docs/user-guide/showcase-walkthrough.md index 62ea3912..0f63837e 100644 --- a/docs/user-guide/showcase-walkthrough.md +++ b/docs/user-guide/showcase-walkthrough.md @@ -156,7 +156,7 @@ run. The Inspect button deep-links to `/chat` where the approved tool call is visible in the transcript. See [Agents and RAG Guide](./agents-and-rag-guide.md) for the approval gate. - +![Agents (HITL) phase — the agent_hitl_flow step card. In this local run the agent answered without calling save_scenario, so the step is skipped by design; the one-click Approve button renders only when the agent triggers the gated save_scenario tool.](./img/agent-hitl-approve.png) ### Phase: Ops snapshot @@ -168,7 +168,7 @@ endpoints are 200-safe on an empty DB, so the step always reports `pass` unless every endpoint fails (then `warn`). The Inspect button deep-links to `/ops`. - +![Ops snapshot step card — the five-tile KPI grid (stale aliases, retraining queue, total runs, total aliases, degrading-health grains) rendered after a green showcase_rich run.](./img/ops-snapshot-tiles.png) ### Cross-cutting polish @@ -198,9 +198,9 @@ Five chrome-level additions wrap the page: phase. Prior behavior pinned the open panel to the running/fallback phase. - - - +![Showcase KPI strip — five tiles populated after a completed run: runs registered, aliases live, batch items completed, scenario plans saved, RAG chunks indexed.](./img/kpi-strip.png) +![Inspect-Artifacts panel — the grid of deep-link cards into every dashboard page populated by the run.](./img/inspect-artifacts-panel.png) +![Run history strip — the most recent runs with scenario, status, wall-clock, and a one-click Replay button.](./img/run-history-strip.png) ## Performance budgets