# Changelog

## Score

- CAI 47 → 54 (+7.1)
- Rubric changed (rubric-2026.10.5 → rubric-2026.11.2) — scores are not directly comparable.

## Lenses

- Code Health 86 → 86 (-0.0)
- Architecture 47 → 47 (-0.6)
- Maturity 38 → 65 (+26.9)
- Readiness 67 → 69 (+2.0)
- Security 49 → 49 (-0.3)
- Performance 77 → 89 (+11.5)

## Resolved (8)

- Consequences section mentions independent evolution but does not state trade-offs such as the runtime binary being a dependency of the CLI or how version mismatches are resolved (docs/decision_records/cli/CLI-001-cli-and-runtime-versioning.md)
- Consequences/trade-offs are thin and trade-off framing is absent: the decision groups APIs under a single namespace/package but does not say whether this increases coupling or reduces fragmentation (docs/decision_records/api/API-003-messaging-api-names.md)
- Context is a single sentence ('We reviewed storage API design for completeness and consistency.') with no problem statement and consequences are restated in the decisions section (docs/decision_records/api/API-001-state-store-api-design.md)
- Context is a single word ('We reviewed parity of state store APIs .') and consequences rest on future API change speculation (Bulk Delete API 'might come in future versions based on the scenarios'), with no trade-offs or rationale for continuing the SaveState API as-is (docs/decision_records/api/API-011-state-store-api-parity.md)
- Context is thin ("Authenticode signing of binaries."), decisions are one sentence with no trade-offs ('In future we can revisit...'), and the decision "Binaries will not be signed with Microsoft keys. In future we can revisit to sign the binaries with dapr.io keys." is a restatement rather than an explicit decision (docs/decision_records/engineering/ENG-004-signing.md)
- Coverage not measured — no coverage collector is wired up
- No consequences/trade-offs section and no trade-off rationale for choosing fasthttp over net/http (docs/decision_records/architecture/ARC-004-http-server.md)
- The concept of the year the car was manufactured is named inconsistently within the same type. One property uses 'ModelYear' while another uses 'BuildYear'.

## New (1)

- Medium CVE: GO-2026-6888 (go.mod)
