Cockpit D2 — evidence freshness projection

[LogicDuke]

Cockpit D2 — evidence freshness projection

DRAFT. No Ready authority. No merge authority. No reviewer triggers.

Baseline

• Base SHA: e1b1a2e5ab2422fbdddc63a7364a9b71bfc0e882 (origin/main, PR PR004-F1 — isolate evidence freshness append descriptors #48 merge commit)
• Base tree: 28e06049a97210acb6edeb5acb0003340b3d26c3
• Head commit: 140e6225f61d4c8192de2820eb0a26a33c44c674 (tree e3b8c4716a554f055262316ce111efa70691444d)

D2 single responsibility

Project PR 004 evidence-freshness answers for the evidence records contained in one already-validated CockpitSnapshot. Presentation / observability only.

Trust boundary — Option A

The public API consumes a CockpitSnapshot that has already passed D1's readCockpitSnapshot() boundary. D1 owns hostile unknown input; D2 adds no second reader, no invalidFields envelope, no malformed-input pseudo-validation. A zero-result projection means the valid snapshot's evidence list is actually empty.

One-way data flow

validated CockpitSnapshot
  -> snapshot evidence read models
  -> minimal EvidenceRecord reconstruction
  -> EvidenceTarget derived from the enclosing snapshot
  -> PR 004 evaluateEvidenceSet()
  -> immutable Cockpit presentation projection

No reverse arrow.

Freshness kernel reuse (PR 004)

EvidenceRecord / EvidenceKind / EvidenceSource from src/domain/evidence.ts; EvidenceTarget / evaluateEvidenceSet / FreshnessState / FreshnessReason / FRESHNESS from src/domain/evidence-freshness.ts. No SHA comparison in Cockpit; state / reason / invalidFields copied verbatim.

Identity rules

• repositoryId injection: every reconstructed record takes repositoryId from snapshot.repository.repositoryId (a D1 snapshot describes exactly one repository; the D1 read model is unchanged).
• observedHeadSha target: the single EvidenceTarget is { repositoryId: snapshot.repository.repositoryId, currentHeadSha: snapshot.repository.observedHeadSha }, read once into locals. No evidence commitSha, advisory echo, finding, PR, reviewer, or metadata value can become HEAD.
• Finding freshness: OUT OF SCOPE. findings[*].advisoryFreshness is never read.

Public API

projectCockpitEvidenceFreshness(snapshot: CockpitSnapshot): CockpitEvidenceFreshnessProjection
// types: CockpitEvidenceFreshnessItem, CockpitEvidenceFreshnessCounts, CockpitEvidenceFreshnessProjection

Flat, input-ordered results + counts { current, stale, invalid, total }. No buckets. Deeply frozen, detached, JSON-round-trip stable, ambient-realm robust (captured intrinsics, null-proto records, insulated descriptors, list toJSON shadow).

Exact file scope (5 files)

• ADD docs/architecture/D2-cockpit-evidence-freshness-projection.md
• ADD src/cockpit/evidence-freshness-projection.ts
• MODIFY src/cockpit/index.ts
• ADD tests/cockpit/evidence-freshness-projection.test.ts
• MODIFY tests/cockpit/architecture-invariants.test.ts (exact single-name exception for projectCockpitEvidenceFreshness; no general project* grant; authority-name checks unchanged)

No D1 read-model modification. No domain kernel modification. No dependency change. No authority expansion. No new bound beyond D1's 1,000 evidence records.

Deterministic verification

• D2 focused: 24 passed
• Cockpit: 4 files / 83 passed
• Domain evidence-freshness: 2 files / 169 passed
• Full suite: 19 files / 989 passed
• typecheck: clean · lint: clean · build: clean · git diff --check: clean

Independent checks

• Fresh independent validator (new claude -p process, no --resume, on the uncommitted candidate): VALIDATION = PASS (7 non-blocking observations; one doc wording note applied before commit)
• Second-order closure: PASS

Untouched

• Deferred findings: untouched
• PR PR 007 — Autoflow State Machine: commit-bound, pure orchestration state #9: untouched
• PR PR004-F1 — isolate evidence freshness append descriptors #48 artifacts: untouched

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 6ca1e274-06e8-4a9b-b4de-69ea04ec675d

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}