Reframe Cockpit as Goal Command Center

docs/cockpit-qa.md

@@ -9,13 +9,26 @@ Docker deployment is intentionally a second pass. This checklist covers local an
    AGENTFLOW_REPOSITORIES=smota/agentflow-sdlc GITHUB_TOKEN=<read-token> pnpm cockpit
    ```
 2. Open `/` and verify Goal Board loads.
+   - View tabs include Goals, Releases, Reviews, and Follow-ups.
+   - Release dashboard is first-class, above the goal workspace.
+   - Release metrics include awaiting release, missing release notes, and release blockers.
 3. Open `/issues/119` and verify:
-   - SDLC timeline renders.
-   - Evidence Health renders.
-   - Relationship Graph renders nodes/edges.
-   - Comment lanes render workflow sections.
+   - Goal detail has a back link to Goal Command Center.
+   - Compact goal header uses stable grammar: identity, status/path/release badges, then actions.
+   - Source is an external header action, not mixed into conceptual badges.
+   - Selected Path and Version / Release Lens appear in the summary strip.
+   - Human review status is explicit when required by high-assurance path.
+   - Next-best-actions render as compact row cards.
+   - Role Flow Contributions render icon/status first instead of generic timeline framing.
+   - Readiness Health renders icon/color state markers, score, denominator, excluded skipped/not-applicable checks, and hover/focus detail.
+   - Relationship Map renders only when relationships exist and explains its navigation purpose.
+   - Activity & Decisions hides empty lanes or shows a compact empty state.
    - Guarded actions panel says server-side auth/CSRF/audit required.
-4. Open `/issues/127/replay` and verify compact Goal Story replay.
+4. Open `/issues/127/replay` and verify Goal Story replay.
+   - Navigation links back to Goal detail and Goal Command Center.
+   - Timeline clearly says `Oldest → newest`.
+   - Events render on a vertical flow line ending at Current state.
+   - Page copy says how the goal moved from intent to current state.
 5. Open `/issues/127/replay.md` and verify markdown export.
 6. Confirm replay page says read-only reconstruction and has no rerun/execute control.
 7. Confirm no raw transcript, prompt, tool input, secret, or full log content appears.
@@ -81,6 +94,20 @@ Use a process manager such as systemd, pm2, or a platform process runner. Termin
 - Disable telemetry with `COCKPIT_RUNNER_TELEMETRY=false`.
 - Remove OAuth app credentials if compromised.
 
+## Goal Command Center UX checklist
+
+- [ ] Header uses repo-local AgentFlow logo and product palette.
+- [ ] Home page answers what goal needs attention and why within one screen.
+- [ ] Top metrics, highlight zone, and next-best-actions are visible above the goal list.
+- [ ] GitHub issue/PR/comment language is secondary to goal, confidence, evidence, next action, and follow-ups.
+- [ ] Fail-safe states avoid generic `unknown` warnings except real errors.
+- [ ] Header hides local token implementation detail and labels the repository as Workspace.
+- [ ] Goal Story appears as a styled action, not an unstyled utility link.
+- [ ] Readiness Health behaves like a health check with score, dimensions, denominator, and floating details.
+- [ ] Skipped-by-path roles explain why they were skipped and do not reduce score.
+- [ ] Release dashboard is a first-class view with release candidate, target branch, awaiting-release goals, missing notes, and blockers.
+- [ ] Version / Release Lens distinguishes merged, awaiting release, delivered/closed, and no-release-impact states.
+
 ## Release checklist
 
 - [ ] `pnpm test` passed.

docs/cockpit.md

@@ -1,6 +1,6 @@
 # AgentFlow Cockpit
 
-AgentFlow Cockpit is the planned optional UI for goal-oriented SDLC delivery. It presents GitHub issues, epics, comments, PRs, role-pass evidence, validation, and follow-ups as AgentFlow workflow state instead of a generic issue list.
+AgentFlow Cockpit is the optional Goal Command Center for goal-oriented SDLC delivery. It turns GitHub issues, epics, comments, PRs, role-pass evidence, validation, and follow-ups into goals, evidence health, role flow contributions, graph navigation, next-best-actions, and replayable goal stories.
 
 Cockpit is optional. The CLI/GitHub workflow remains authoritative and fully usable without Cockpit.
 
@@ -45,17 +45,17 @@ Cockpit does not require a mounted repo checkout in remote mode. Future Docker d
 
 ## Opinionated views
 
-### Goal Board
+### Goal Command Center
 
-Shows goals grouped by workflow state: Intake, Scoped, Planned, Implementation, Validation, Review, PR Ready, Done, and Blocked.
+Shows top metrics, a highlight goal, next-best-actions, goal health, follow-ups, and human gates. GitHub issue details are secondary metadata; the primary language is goal, confidence, evidence, next action, and role flow.
 
 ### Epic View
 
 Rolls up child issues, dependencies, acceptance matrix, risk, active workstreams, PRs, follow-ups, and decisions.
 
-### Issue SDLC View
+### Goal Detail, Selected Path, and Role Flow Contributions
 
-Shows phases 0-8:
+Shows the selected path/profile/risk first, then phases 0-8 as role contributions rather than a generic timeline. Roles can be complete, skipped by path, not applicable, not started, needs attention, blocked, or error:
 
 1. Product manager / JTBD
 2. Analyst
@@ -67,20 +67,24 @@ Shows phases 0-8:
 8. Tech writer
 9. PR readiness
 
-Each phase shows role-pass completeness, evidence, validation, handover, and next role contract.
+Each role card shows completion first. Supporting evidence is detail/proof behind the role status. Roles skipped by selected path explain why and do not reduce readiness.
 
-### Evidence Health
+### Readiness Health
 
-Reports governance completeness, not model confidence:
+Reports governance completeness as a path-aware score with explainable dimensions, not model confidence. The denominator includes only applicable checks; skipped-by-path and not-applicable checks are excluded:
 
-- Required policy read state.
-- Acceptance criteria.
-- Role-pass fields.
-- Validation status.
-- Review independence.
-- Human gate requirements.
-- PR manifest readiness.
-- Follow-up disposition.
+- Scope.
+- Design.
+- Implementation.
+- Validation.
+- Review/gates.
+- Follow-ups.
+
+Missing optional evidence is shown with fail-safe copy instead of generic `unknown` warnings.
+
+### Version / Release Lens
+
+Shows whether the goal is in progress, merged, awaiting release, or delivered/closed. It also surfaces target branch, release impact, candidate version when known, and release-note state. This reinforces that merged work is not always delivered work.
 
 ### Comments lanes
 

lib/__tests__/cockpit-goal-model.test.mjs

@@ -0,0 +1,121 @@
+import { describe, expect, it } from 'vitest'
+import {
+  buildAgentFlowGoalModel,
+  buildCommandCenterModel,
+  buildReleaseDashboard,
+  deriveHumanGate,
+  deriveSelectedPath,
+  deriveVersionLens,
+  DISPLAY_STATES,
+} from '../cockpit-goal-model.mjs'
+
+describe('AgentFlow goal model', () => {
+  it('derives health score and next-best-actions from GitHub evidence', () => {
+    const goal = buildAgentFlowGoalModel({
+      issue: issue({ body: '## Acceptance criteria\n- [ ] works\n\n## Test plan\n- pnpm test' }),
+      comments: [{ body: '<!-- agentflow:validation-summary -->\npnpm test passed' }],
+      pullRequests: [{ number: 1, title: 'PR', merged_at: null }],
+    })
+    expect(goal.evidenceHealth.score).toBeGreaterThan(60)
+    expect(goal.nextBestActions.map((action) => action.id)).toContain('request-review')
+    expect(goal.roleContributions.find((item) => item.phase === 'analyst').state).toBe(
+      DISPLAY_STATES.recorded,
+    )
+  })
+
+  it('prioritizes human gate for sensitive goals', () => {
+    const goal = buildAgentFlowGoalModel({
+      issue: issue({
+        body: '## Acceptance criteria\n- [ ] works\n\nRemote auth security high-assurance',
+      }),
+    })
+    expect(goal.nextBestActions[0].id).toBe('request-human-gate')
+    expect(goal.status).toBe('blocked')
+  })
+
+  it('builds command center metrics and highlight', () => {
+    const model = buildCommandCenterModel({
+      issues: [issue({ number: 1, title: 'A' }), issue({ number: 2, title: 'B', body: '' })],
+    })
+    expect(model.metrics.activeGoals).toBe(2)
+    expect(model.topActions.length).toBeGreaterThan(0)
+    expect(model.highlight).toBeTruthy()
+  })
+
+  it('excludes skipped-by-path roles from readiness denominator', () => {
+    const goal = buildAgentFlowGoalModel({
+      issue: issue({
+        body: '## Acceptance criteria\n- [ ] typo fixed\n\n## Test plan\n- docs reviewed\n\ndocs-only bounded low-risk',
+      }),
+      comments: [{ body: 'validation passed. review complete.' }],
+      pullRequests: [{ number: 2, title: 'Docs PR', base: { ref: 'development' } }],
+    })
+    expect(goal.selectedPath.profile).toBe('bounded')
+    expect(goal.roleContributions.find((item) => item.phase === 'architect')).toMatchObject({
+      status: 'skipped-by-path',
+      scoreImpact: 'excluded',
+    })
+    expect(goal.evidenceHealth.excluded).not.toContain('roleFlow')
+    expect(goal.evidenceHealth.dimensions.find((item) => item.id === 'roleFlow').detail).toContain(
+      'applicable roles',
+    )
+  })
+
+  it('builds first-class release dashboard rollup', () => {
+    const model = buildCommandCenterModel({
+      issues: [issue({ number: 1, title: 'Ship A', labels: [{ name: 'feature' }] })],
+      pullRequestsByIssue: {
+        1: [{ number: 3, merged_at: '2026-01-01T00:00:00Z', base: { ref: 'development' } }],
+      },
+    })
+    expect(model.releaseDashboard).toMatchObject({ targetBranch: 'development' })
+    expect(model.releaseDashboard.awaitingRelease).toHaveLength(1)
+    expect(model.metrics.missingReleaseNotes).toBe(1)
+    expect(buildReleaseDashboard(model.goals).includedGoals).toHaveLength(1)
+  })
+
+  it('derives version lens for merged development work awaiting release', () => {
+    const version = deriveVersionLens({
+      issue: issue({ labels: [{ name: 'feature' }] }),
+      pullRequests: [
+        { number: 3, merged_at: '2026-01-01T00:00:00Z', base: { ref: 'development' } },
+      ],
+    })
+    expect(version.state).toBe('awaiting-release')
+    expect(version.releaseImpact).toBe('minor')
+    expect(version.releaseNoteState).toBe('needs-decision')
+  })
+
+  it('infers high-assurance selected path from sensitive work', () => {
+    expect(
+      deriveSelectedPath({
+        issue: issue({ body: '## Acceptance criteria\n- [ ] secure\n\nRemote auth security' }),
+      }),
+    ).toMatchObject({ profile: 'high-assurance', risk: 'high' })
+  })
+
+  it('tracks required human gate status explicitly', () => {
+    const selectedPath = deriveSelectedPath({
+      issue: issue({ body: '## Acceptance criteria\n- [ ] secure\n\nRemote auth security' }),
+    })
+    expect(deriveHumanGate({ selectedPath, comments: [] })).toMatchObject({
+      required: true,
+      status: 'not-requested',
+    })
+    expect(
+      deriveHumanGate({
+        selectedPath,
+        comments: [{ body: 'human security review approved and complete' }],
+      }),
+    ).toMatchObject({ status: 'complete' })
+  })
+})
+
+function issue({
+  number = 1,
+  title = 'Goal',
+  body = '## Acceptance criteria\n- [ ] works',
+  labels = [{ name: 'drafted-by:pi' }],
+} = {}) {
+  return { number, title, body, labels, state: 'open' }
+}

lib/__tests__/cockpit-read-model.test.mjs

@@ -11,17 +11,20 @@ describe('cockpit read model', () => {
     expect(view.classification.isAgentFlowManaged).toBe(true)
     expect(view.commentLanes.Handover).toHaveLength(1)
     expect(view.evidenceHealth.status).toBe('complete')
+    expect(view.goal.title).toBe('Read-only MVP')
+    expect(view.goal.nextBestActions.length).toBeGreaterThan(0)
   })
 
-  it('builds goal board columns with blocked evidence surfaced', () => {
+  it('builds command center metrics with attention surfaced', () => {
     const board = buildGoalBoard({
       issues: [
         managedIssue({ number: 1, title: 'Good' }),
         { number: 2, title: 'Missing', labels: [], body: '' },
       ],
     })
-    expect(board.Scoped).toHaveLength(1)
-    expect(board.Blocked).toHaveLength(1)
+    expect(board.metrics.activeGoals).toBe(2)
+    expect(board.goals).toHaveLength(2)
+    expect(board.topActions.length).toBeGreaterThan(0)
   })
 
   it('rolls up epic child evidence', () => {
@@ -37,7 +40,7 @@ describe('cockpit read model', () => {
       ],
     })
     expect(rollup.totals.children).toBe(2)
-    expect(rollup.totals.blocked).toBe(1)
+    expect(rollup.totals.blocked).toBe(2)
     expect(rollup.totals.evidenceComplete).toBe(1)
   })
 })