@surlymochan/evalid 0.5.3 → 0.5.4

package/CHANGELOG.md

@@ -5,6 +5,13 @@ All notable changes to this project are documented here. Detailed notes live in
 
 ## [Unreleased]
 
+## [0.5.4] - 2026-11-07
+
+### Fixed
+- **Phase-cost drift test stops flaking under load (#44, issue #5 P1-1)**: the two `within_tolerance` integration tests used 2 x 400 ms sleeps, so the 5% drift budget was 40 ms against tens of ms of fixed executor overhead outside phases (run setup plus `phase_completed` log / state writes). On main this tripped at 5.7% drift while the MiniMax batch was running, then the same file went 7/7 green on a quiet re-run — a load-dependent flake at a marginal threshold, not an accounting bug (the accounting correctly flagged the drift). The sleeps are now 2 x 1500 ms, putting the same overhead at ~2% of the phase sum with headroom under load. Production tolerance and accounting unchanged. Test evidence: `node test/phase-cost.test.js` 7/7 green; full `npm run verify` exit 0 (43 files, 361 tests, 360 pass, 1 skipped pre-existing, 0 fail).
+
+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 running — no bench numbers are claimed in this release.
+
 ## [0.5.3] - 2026-11-07
 
 ### Fixed

package/RELEASE_NOTES.md

@@ -1,5 +1,13 @@
 # Release Notes
 
+## 0.5.4 - Phase-cost drift test stops flaking under load
+
+Patch release. One test-only fix: `npm run verify` was red on main because the phase-cost drift test tripped its 5% bound at 5.7% while the MiniMax batch was running, then went 7/7 green on a quiet re-run of the same file.
+
+**Longer sleeps in the drift tests (#44, issue #5 P1-1).** The two `within_tolerance` integration tests used 2 x 400 ms sleeps, so the 5% budget was 40 ms against tens of ms of fixed executor overhead that structurally sits outside phases (run setup plus the `phase_completed` log / state writes between phases). The sleeps are now 2 x 1500 ms, putting the same overhead at ~2% of the phase sum with headroom under load. The production tolerance (`PHASE_COST_TOLERANCE_PCT = 5`) and the accounting in `src/core/phase-cost.js` / `workflow-executor.js` are unchanged — the accounting behaved correctly by flagging real overhead. Test evidence (from the PR): `node test/phase-cost.test.js` 7/7 green; full `npm run verify` exit 0 (43 files, 361 tests, 360 pass, 1 skipped pre-existing, 0 fail).
+
+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 running. No bench numbers are claimed in this release.
+
 ## 0.5.3 - Bench planner stops timing out and stops wandering
 
 Patch release. Two bench-harness fixes from the MiniMax pilot, both pinned by tests rather than by a live run.

package/docs/launch-dossier.md

@@ -1,6 +1,6 @@
 # evalid Launch Dossier
 
-Generated: 2026-11-07T05:49:01.426Z
+Generated: 2026-11-07T07:29:35.000Z
 Mode: post_publish
 Verdict: conditionally ready: engine and adoption evidence are strong, registry publication remains required for a splash launch
 Launch ready: not yet
@@ -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@0.5.3
+- Package name/version: @surlymochan/evalid@0.5.4
 
 ## Passing Evidence
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@surlymochan/evalid",
-  "version": "0.5.3",
+  "version": "0.5.4",
   "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",