@surlymochan/evalid 0.5.10 → 0.5.12

package/CHANGELOG.md

@@ -5,6 +5,21 @@ All notable changes to this project are documented here. Detailed notes live in
 
 ## [Unreleased]
 
+## [0.5.12] - 2026-11-08
+
+### Added
+- **Policy resolve atom coverage (#66)**: `src/core/atoms/policy/resolve.js` (P1 `policy.resolve`) was at 0% coverage. Added 6 tests to `test/policy-overlay.test.js` covering personal-default resolution with `policy_effective.json` writing, change_id and execution_mode taken from input, `team_policy_missing` / `team_policy_invalid` passthrough, no artifact when no change id is supplied, and registry wiring. Test evidence: RED→GREEN (stubbed-throwing implementation failed the new tests, then green after restore); `node --test test/policy-overlay.test.js` 18 pass / 0 fail; `npm test` 435 tests / 434 pass / 0 fail / 1 skipped (pre-existing); coverage after: resolve.js 100% statements / 83.33% branches / 100% functions / 100% lines (was 0% before; the single uncovered branch is the `process.cwd()` fallback, untested by design to avoid polluting the repo).
+
+Acceptance debt, stated plainly: the bench-number acceptance for P1-2 / P1-3 / P2-1 / P2-3 still awaits the MiniMax full-matrix batch (15 instances × 4 arms × reps 3), which is stopped on MiniMax Token Plan quota exhaustion. No bench numbers are claimed in this release.
+
+## [0.5.11] - 2026-11-08
+
+### Added
+- **Codex app bridge atom coverage (#63)**: `src/agent_team/atoms/codex-app-bridge.js` (I5 `codex.app_bridge`) was at 0% coverage. Added 6 tests to `test/heavy-atoms.test.js` covering the default local-bridge placeholder return, input.phase preference over context.phase.id, empty-string and missing-phase fallbacks, and registry wiring. Test evidence: RED→GREEN (stubbed-throwing implementation failed all 6 new tests, then green after restore); `npm run coverage` 424 pass / 0 fail; coverage after: codex-app-bridge 100% (all four measures, was 0% before).
+- **Visual layout competition atom coverage (#64)**: `src/agent_team/atoms/visual/layout-competition.js` (H3b `visual.layout_competition`) was at 0% coverage. Added 4 tests to `test/heavy-atoms.test.js` covering `layout_competition.json` writing with the fixed phase and summary (verdict accept / confidence 0.76, two findings), the fixed phase winning over input/context phase names, the missing-change_id error, and registry wiring. Test evidence: RED→GREEN (an intentionally wrong assertion failed, then green after restore); `node test/heavy-atoms.test.js` 64 pass / 0 fail; `npm run coverage` 429 tests / 428 pass / 0 fail / 1 skipped (pre-existing); coverage after: layout-competition 100 / 100 / 100 / 100 (was 0% before).
+
+Acceptance debt, stated plainly: the bench-number acceptance for P1-2 / P1-3 / P2-1 / P2-3 still awaits the MiniMax full-matrix batch (15 instances × 4 arms × reps 3), which is stopped on MiniMax Token Plan quota exhaustion. No bench numbers are claimed in this release.
+
 ## [0.5.10] - 2026-11-08
 
 ### Added

package/RELEASE_NOTES.md

@@ -1,5 +1,23 @@
 # Release Notes
 
+## 0.5.12 - Policy resolve atom test coverage
+
+Patch release. One test-coverage addition since 0.5.11, verified by the full test suite. No runtime behaviour changed.
+
+**Policy resolve atom coverage (#66).** P1 `policy.resolve` (`src/core/atoms/policy/resolve.js`) now has 6 tests in `test/policy-overlay.test.js`, covering personal-default resolution with `policy_effective.json` writing, change_id and execution_mode taken from input, `team_policy_missing` / `team_policy_invalid` passthrough, no artifact when no change id is supplied, and registry wiring. Coverage went from 0% to 100% statements / 83.33% branches / 100% functions / 100% lines (the single uncovered branch is the `process.cwd()` fallback, untested by design). Test evidence: RED→GREEN verified (stubbed-throwing implementation failed the new tests, then green after restore); `npm test` 435 tests / 434 pass / 0 fail / 1 skipped (pre-existing).
+
+Acceptance debt, stated plainly: the bench-number acceptance for P1-2, P1-3, P2-1 and P2-3 — the wall-clock/token/resolved lines from issue #5 — still awaits the MiniMax full-matrix batch (15 instances × 4 arms × reps 3), which is stopped on MiniMax Token Plan quota exhaustion. No bench numbers are claimed in this release.
+
+## 0.5.11 - Codex app bridge and layout competition atoms test coverage
+
+Patch release. Two test-coverage additions since 0.5.10, both verified by the full test suite. No runtime behaviour changed.
+
+**Codex app bridge atom coverage (#63).** I5 `codex.app_bridge` (`src/agent_team/atoms/codex-app-bridge.js`) now has 6 tests in `test/heavy-atoms.test.js`, covering the default local-bridge placeholder return, input.phase preference over context.phase.id, empty-string and missing-phase fallbacks, and registry wiring. Coverage went from 0% to 100% on all four measures. Test evidence: RED→GREEN verified (stubbed-throwing implementation failed all 6 new tests, then green after restore); `npm run coverage` 424 pass / 0 fail.
+
+**Visual layout competition atom coverage (#64).** H3b `visual.layout_competition` (`src/agent_team/atoms/visual/layout-competition.js`) now has 4 tests in `test/heavy-atoms.test.js`, covering `layout_competition.json` writing with the fixed phase and summary (verdict accept / confidence 0.76, two findings), the fixed phase winning over input/context phase names, the missing-change_id error, and registry wiring. Coverage went from 0% to 100 / 100 / 100 / 100. Test evidence: RED→GREEN verified (an intentionally wrong assertion failed, then green after restore); `node test/heavy-atoms.test.js` 64 pass / 0 fail; `npm run coverage` 429 tests / 428 pass / 0 fail / 1 skipped (pre-existing).
+
+Acceptance debt, stated plainly: the bench-number acceptance for P1-2, P1-3, P2-1 and P2-3 — the wall-clock/token/resolved lines from issue #5 — still awaits the MiniMax full-matrix batch (15 instances × 4 arms × reps 3), which is stopped on MiniMax Token Plan quota exhaustion. No bench numbers are claimed in this release.
+
 ## 0.5.10 - Intake and phase-run fail atoms test coverage
 
 Patch release. Two test-coverage additions since 0.5.9, both verified by the full test suite. No runtime behaviour changed.

package/docs/launch-dossier.md

@@ -117,7 +117,7 @@ Evidence:
 - Adoption records: 1; status: pass
 - Goal audit: pass; goal: Keep evalid credible as an open-source agent delivery runtime: repo-owned workflow contracts, executable verification gates, durable project context, and honest benchmark evidence.
 - Package evidence: missing; source: registry_json
-- Package name/version: @surlymochan/evalid@0.5.10
+- Package name/version: @surlymochan/evalid@0.5.12
 
 ## Passing Evidence
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@surlymochan/evalid",
-  "version": "0.5.10",
+  "version": "0.5.12",
   "description": "Claimed != verified: a runtime verification layer for coding agents. It seals each test run and replays it from a clean snapshot, with a zero-judge gate pipeline (spec clarification, code review, TDD proof) and the xflow CLI.",
   "keywords": [
     "ai-agents",