diff --git a/agentops-cli/src/commands/dashboard.js b/agentops-cli/src/commands/dashboard.js index 4348a9f..daa7837 100644 --- a/agentops-cli/src/commands/dashboard.js +++ b/agentops-cli/src/commands/dashboard.js @@ -561,6 +561,7 @@ function validateDashboardUx() { schema_version_coverage: true, exporter_failure_review: true, pattern_drilldowns: true, + run_centric_ui: true, empty_state_dashboards: emptyStateDashboards }, errors diff --git a/agentops-cli/src/commands/product.js b/agentops-cli/src/commands/product.js index 20aa2bf..4358155 100644 --- a/agentops-cli/src/commands/product.js +++ b/agentops-cli/src/commands/product.js @@ -209,6 +209,24 @@ function productAudit(options = {}) { checks.push(check('dashboard-drilldowns', links.ok, [`${links.checked_links} nav/data links checked`], links.errors)); const ux = validateDashboardUx(); checks.push(check('dashboard-operator-ux', ux.ok, ['Home, Runs, Replay, transcript, patterns, recommendations, and empty states checked'], ux.errors)); + checks.push(check( + 'run-centric-ui-contract', + ux.ok + && ux.contracts?.run_centric_ui === true + && fileIncludes('docs/agentops-architecture-product-audit.md', [ + 'Run-Centric UI', + 'Session explorer as first screen', + 'Trace waterfall through Run Replay', + 'Ask AgentOps panel' + ]), + [ + 'grafana/dashboards/v2/01-agentops-home.json', + 'grafana/dashboards/v2/02-runs-explorer.json', + 'grafana/dashboards/v2/03-run-replay.json', + 'docs/agentops-architecture-product-audit.md' + ], + ux.errors + )); checks.push(check( 'azure-ingest-privacy-plan', diff --git a/agentops-cli/test/index.test.js b/agentops-cli/test/index.test.js index fa1ceeb..988e73c 100644 --- a/agentops-cli/test/index.test.js +++ b/agentops-cli/test/index.test.js @@ -4331,6 +4331,7 @@ test('product audit proves the local AgentOps control-room contract', () => { 'evals-insights-recommendations', 'grafana-v2-pack', 'kql-library', + 'run-centric-ui-contract', 'content-transcript-opt-in', 'first-run-loop', 'ask-agentops-response-flow' diff --git a/docs/agentops-architecture-product-audit.md b/docs/agentops-architecture-product-audit.md index fe0f4b4..cb089fa 100644 --- a/docs/agentops-architecture-product-audit.md +++ b/docs/agentops-architecture-product-audit.md @@ -1203,16 +1203,17 @@ Success criterion: The default UI answers: what happened, why did it fail or cost money, what changed, what should I do next? ``` -Required work: +Implemented: -- Session explorer as first screen. -- Trace waterfall. -- Policy/safety strip. -- Tool/MCP waterfall. -- Context/tokens panel. -- Recommendation panel. -- Eval/benchmark linkage. -- Ask AgentOps panel. +- Session explorer as first screen through the Home **Session Health** panel and the **Runs Explorer** dashboard. +- Trace waterfall through Run Replay's **Replay timeline** and `OpenTrace` drilldowns. +- Policy/safety strip through Run Replay's **Policy, privacy, tests, and GitHub outcome** panel plus the Safety, Privacy & Policy dashboard. +- Tool/MCP waterfall through Run Replay's **Agent, skill, and MCP lineage** panel plus the Tools & MCP Risk dashboard. +- Context/tokens panel through Run Replay's **Context and cache posture** panel and the Models, Cost & Tokens dashboard. +- Recommendation panel through Home **Recommended next actions**, Run Replay **Latest recommendation**, and Insights **Recommendation artifacts**. +- Eval/benchmark linkage through Evals & Quality scorecards, before/after comparison, artifact review, hidden checks, policy review, semantic checks, and promotion approvals. +- Ask AgentOps panel through Run Replay **Ask AgentOps context** and hosted Ask AgentOps action links. +- `agentops product audit` now includes `run-centric-ui-contract` so this UI shape remains a local release gate. ### 3. Agent Improvement Loop