v22.14.0
Release class: binary-release — this release changes the compiled binary, so it has a NEW macOS fingerprint that needs a corporate approval request (see the Fingerprints section below).
Generated by evolve release (go/internal/changeloggen) from commits v22.13.1..HEAD.
Added
- repo-contract scanner pack — lane ships can no longer red main (#413)
Fixed
- restore incident-postmortem to its durable spec — cycle-1313's ship reworked the phase config against the test-pinned domain-phase-catalog contract (sections renamed off-spec; persona declared FOREIGN tool vocabulary view_file/grep_search/run_command vs the Claude-vocabulary profile; novel kind field). Per-cycle scope never ran the repo-wide catalog tests (scope-disease costume #5) so main went red on the merge run + the persona-coherence ship gate blocked plane ships. Repair-in-place (the same ship carries the legit F4 triage admission — no revert): spec sections Impact/Timeline/Root Cause/Action Items restored, persona tools aligned to profile vocabulary, kind dropped, stale allowlist entry deleted (ratchet shrinks). phasespec + phasecoherence green. (#412)
- boot-time binary staleness self-heal (binary-lag class) — twice-adversarially-reviewed (#411)
Documentation
- engineering chronicle (17 narratives) + doc-root consolidation into docs/ (#410)
- batch integrity review 2026-08-04 — six findings in issue/gap/solution format (defect laundering across salvage chains incl. the 1255-D1 stale-worktree CRITICAL narrowed to 'verified closed'; TIA dormant-code-under-active-soak status fiction with vacuous soak bar; audit-WARN prescriptions unenforced at ship (1258 durable eval exists nowhere, predicate green-by-skip); triage protected-surface admission gap that burned 3 lanes; dead-red predicate corpus; ship-claim misattribution). Verdict tables for all 9 ships + 7 FAILs (code SUBSTANTIVE, FAILs honest, gaming lives in status accounting). TDD/design conformance follow-up for #398-#407. Operating-policy additions: §3.8 issue/gap/solution docs on every fix (operator directive) + §3.9 ledger-writes-derive-from-diffs with per-defect continuation dispositions. Queue reprioritized: the five review items hold ranks 1/2/4/5 (0.96/0.95/0.91/0.90). (#408)
Other
- Merge remote-tracking branch 'origin/main'
- (dossier) cycle-1320 closeout
- (dossier) cycle-1319 closeout
- (dossier) cycle-1318 closeout
- (dossier) cycle-1317 closeout
- (dossier) cycle-1315 closeout
- (dossier) cycle-1316 closeout
- (dossier) cycle-1314 closeout
- (dossier) cycle-1313 closeout
- evolve-cycle: goal=9d22f7f9398fa4e2fc8c2a77c8315a906816e20057ec26c8f437295cf59ca517
- (dossier) cycle-1312 closeout
- (dossier) cycle-1311 closeout
- (dossier) cycle-1309 closeout
- (dossier) cycle-1310 closeout
- (dossier) cycle-1308 closeout
- (dossier) cycle-1307 closeout
- (dossier) cycle-1305 closeout
- (dossier) cycle-1306 closeout
- (dossier) cycle-1303 closeout
- (dossier) cycle-1304 closeout
- (dossier) cycle-1301 closeout
- (dossier) cycle-1302 closeout
- (dossier) cycle-1300 closeout
- (dossier) cycle-1299 closeout
- (dossier) cycle-1298 closeout
- (dossier) cycle-1297 closeout
- (dossier) cycle-1296 closeout
- (dossier) cycle-1295 closeout
- (dossier) cycle-1294 closeout
- (dossier) cycle-1293 closeout
- (dossier) cycle-1291 closeout
- (dossier) cycle-1292 closeout
- (dossier) cycle-1289 closeout
- (dossier) cycle-1290 closeout
- (dossier) cycle-1288 closeout
- (research) deliverable-alignment strategic evaluation 2026-08 — two-track research (local institutional memory + 2025-26 state of the art) answering the operator's find-one-solve-one question. Verdict: right VERIFICATION posture, wrong PREVENTION strategy — the four-layer model assigns each failure class its layer (generation-point / transport+salvage / verification / accounting). Measured record: ~65% of alignment failures were HARNESS defects; contract-tail took bad_verdict to zero; adversarial audit best catch-rate; breaker is a fail-open ratchet under weak CLIs. Ranked 7-move portfolio filed to the live queue: contract-block-cli-escalation 0.96 (rank 1 — refiled: the original existed only in the tracked snapshot, a live instance of the stranding it documents), schema-aligned-salvage-layer 0.9, crossartifact-invariant-stack 0.85, claude-stop-hook-finish-gate 0.8; two-stage verdict minting folds into verdict-sentinel-as-tool-call 0.86. Full citations. (#409)
- (dossier) cycle-1287 closeout
- (dossier) cycle-1286 closeout
- (dossier) cycle-1285 closeout
- (dossier) cycle-1284 closeout
- (dossier) cycle-1282 closeout
- (dossier) cycle-1283 closeout
- (dossier) cycle-1281 closeout
- (dossier) cycle-1280 closeout
- (dossier) cycle-1279 closeout
- (dossier) cycle-1278 closeout
- (dossier) cycle-1277 closeout
- (dossier) cycle-1276 closeout
- (dossier) cycle-1275 closeout
- (dossier) cycle-1274 closeout
- (dossier) cycle-1272 closeout
- (dossier) cycle-1273 closeout
- (dossier) cycle-1270 closeout
- (dossier) cycle-1271 closeout
Fingerprints (corporate approval)
The recommended macOS artifact is the universal evolve_darwin_all.tar.gz (a lipo'd x86_64 + arm64 fat binary) — one fingerprint covering both Intel and Apple Silicon. SHA256 checksums for every published artifact are in the checksums.txt release asset. (The per-arch evolve_darwin_amd64.tar.gz / evolve_darwin_arm64.tar.gz archives remain published for existing installs, each with its own distinct hash — approve the universal one.)
To adopt this release under a fingerprint-based approval system:
- Download
evolve_darwin_all.tar.gzandchecksums.txtfrom this release. - Verify integrity:
shasum -a 256 -c checksums.txt. - Submit the universal binary's SHA256 as the one macOS approval fingerprint (one request per adopted version; the pin re-adopts automatically on first run).