From c21c1d9f5a140f6671ec7910b348606cf4dd5297 Mon Sep 17 00:00:00 2001 From: mzored Date: Thu, 8 Oct 2026 14:35:33 +0400 Subject: [PATCH] Release 6.3.2 with optional contributor-check admission --- CHANGELOG.md | 18 ++++++++++++++ VERSION | 2 +- docs/evidence.md | 28 ++++++++++++++++++++++ evals/cases.json | 2 +- evals/cto-cases.json | 2 +- evals/host-smoke.json | 2 +- plugins/skiphow/.claude-plugin/plugin.json | 2 +- plugins/skiphow/.codex-plugin/plugin.json | 2 +- site/evidence/index.html | 4 ++-- 9 files changed, 54 insertions(+), 8 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 13001ac..5deb79c 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -2,6 +2,24 @@ All notable changes to SkipHow 2.x and later appear in this file. Earlier release notes remain available on [GitHub Releases](https://github.com/mzored/SkipHow/releases). +## 6.3.2 (2026-10-08) + +Contributor checks now use the optional installed `agent-verify` host queue before their subprocess +execution budgets begin. Both deterministic and host verification reuse a validated inherited grant, +keep queue output visible, and stop on adapter errors. Hosts without the capability run directly. +The shared adapter remains outside the shipped plugin and adds no package dependency. + +Independent source review of [PR #143](https://github.com/mzored/SkipHow/pull/143) found no material +integration defects. Deterministic tests cover optional discovery, inherited-grant responses, argument +preservation, admission before execution budgets, and POSIX queue refusal and cancellation on both +entrypoints. Real nested-grant lifecycle, adapter watchdog and admitted-descendant cancellation, +and Windows execution remain `UNVERIFIED`. + +This PATCH publishes contributor-tooling changes. Public skills, authority, installation, and runtime +behavior are unchanged. No paid behavioral experiments or clean-install sessions were run. Current +activation, continuity, and model behavior remain `UNVERIFIED`. +See [current evidence](docs/evidence.md#632-optional-contributor-check-admission). + ## 6.3.1 (2026-10-07) Clarify planning boundaries and share independent-review continuation between plan and longrun. diff --git a/VERSION b/VERSION index dc0208a..91e4a9f 100644 --- a/VERSION +++ b/VERSION @@ -1 +1 @@ -6.3.1 +6.3.2 diff --git a/docs/evidence.md b/docs/evidence.md index a41a57b..d4b7f10 100644 --- a/docs/evidence.md +++ b/docs/evidence.md @@ -2,6 +2,34 @@ This page separates package checks from observed model behavior. The full 2.0 evidence remains in the immutable [`v2.0.1` research snapshot](https://github.com/mzored/SkipHow/tree/1c811262e6acdbdc58a2ee862b54e0b8d3478eaa/docs/research/2026-08-27). +## 6.3.2 optional contributor-check admission + +[PR #143](https://github.com/mzored/SkipHow/pull/143) integrated optional host admission into both +contributor verification entrypoints at `20cbb4461c35f4c0655009ea69400170db649702`. A shared helper +discovers `agent-verify` on PATH, reuses a grant only when its inherited probe succeeds, and otherwise +replaces the checker with the adapter before captured child execution budgets begin. Missing adapters +retain direct execution; probe and startup errors stop the check. The plugin receives no queue code, +dependency, installation change, or runtime-policy change. + +Independent read-only source review on 2026-10-08 compared that revision with the previous release and +checked the installed adapter's protocol. It found no supported material integration defects. The +tests cover absent adapters, inherited-probe responses, startup errors, argument preservation, both +entrypoints' admission boundaries, and real POSIX process replacement, refusal status, and cancellation +while queued. Queue policy, grant validity, watchdogs, and admitted-descendant cancellation belong to +the external adapter. Real nested-grant lifecycle, watchdog and admitted-descendant cancellation, +and Windows execution were not exercised and remain `UNVERIFIED`. + +The maintainer's pinned environment passed the full deterministic gate and `git diff --check` on that +integrated source. Strict Claude schema validation passed; the local Codex validator was unavailable. +[Main CI](https://github.com/mzored/SkipHow/actions/runs/37721276758) passed the deterministic gate and +pinned Codex schema validation for the same revision. Those results establish their named source and +host contracts, not model behavior. The release validation matrix records the final tagged candidate's +capabilities separately. + +Version 6.3.2 aligns release metadata and current-package instruments while preserving the 6.3.1 public +runtime contract. No paid model sessions or clean-install receipts were run. Current installation, +activation, continuity, behavioral improvement, and optimal resource sizing remain `UNVERIFIED`. + ## 6.3.1 planning boundaries and review continuation One private installed Claude Code session on exact 6.3.0 was inspected on 2026-10-07, including subagent briefs, diff --git a/evals/cases.json b/evals/cases.json index 304221d..9f270ea 100644 --- a/evals/cases.json +++ b/evals/cases.json @@ -1,6 +1,6 @@ { "corpus_version": 4, - "package_under_test": "6.3.1", + "package_under_test": "6.3.2", "purpose": "Synthetic cases for three separate instruments: activation, forced-activation CTO behavior, and host smoke. Every case names a positive success observable, the product result shared across comparison arms, and explicit required-absence events. Nothing here has been run.", "not_a_gate": "No model run gates a pull request. python scripts/check.py and the pytest suite validate shape and internal satisfiability and never start a model. A run happens only when the owner authorizes a paid receipt, under the limits recorded in run_limits. A deterministic check passing is never evidence of behavior.", "evidence_labels": { diff --git a/evals/cto-cases.json b/evals/cto-cases.json index 72f1016..a424584 100644 --- a/evals/cto-cases.json +++ b/evals/cto-cases.json @@ -1,6 +1,6 @@ { "instrument": "forced_activation_behavior", - "package_under_test": "6.3.1", + "package_under_test": "6.3.2", "suite_status": "not_run", "minimum_coverage": { "case_ids": [ diff --git a/evals/host-smoke.json b/evals/host-smoke.json index a8dad27..adfdb6e 100644 --- a/evals/host-smoke.json +++ b/evals/host-smoke.json @@ -1,6 +1,6 @@ { "instrument": "host_smoke", - "package_under_test": "6.3.1", + "package_under_test": "6.3.2", "scope": "external_candidate_receipts", "checks": { "clean_install": { diff --git a/plugins/skiphow/.claude-plugin/plugin.json b/plugins/skiphow/.claude-plugin/plugin.json index 84918e7..02f9e5e 100644 --- a/plugins/skiphow/.claude-plugin/plugin.json +++ b/plugins/skiphow/.claude-plugin/plugin.json @@ -1,6 +1,6 @@ { "name": "skiphow", - "version": "6.3.1", + "version": "6.3.2", "description": "Adaptive virtual CTO for founders and product owners using Claude Code or Codex. Describe the product outcome; SkipHow owns the technical lifecycle through verified completion.", "author": { "name": "mzored", diff --git a/plugins/skiphow/.codex-plugin/plugin.json b/plugins/skiphow/.codex-plugin/plugin.json index 9d67392..d8d21ba 100644 --- a/plugins/skiphow/.codex-plugin/plugin.json +++ b/plugins/skiphow/.codex-plugin/plugin.json @@ -1,6 +1,6 @@ { "name": "skiphow", - "version": "6.3.1", + "version": "6.3.2", "description": "Adaptive virtual CTO for founders and product owners using Claude Code or Codex. Describe the product outcome; SkipHow owns the technical lifecycle through verified completion.", "author": { "name": "mzored", diff --git a/site/evidence/index.html b/site/evidence/index.html index 4203e5b..9ddde93 100644 --- a/site/evidence/index.html +++ b/site/evidence/index.html @@ -56,11 +56,11 @@

Claims stop where the receipts stop.

-
+

01 · Historical observations

What controlled 2.x runs showed.

-

Historical observations below: 2.x only. Current package: 6.3.1. Retained current-package CTO scenarios with Observed receipts: 0 of 12. Eight historical 4.0.1 Claude run records remain incomplete. Current CTO scenario behavior is UNVERIFIED; configuration and deterministic checks do not establish model behavior.

+

Historical observations below: 2.x only. Current package: 6.3.2. Retained current-package CTO scenarios with Observed receipts: 0 of 12. Eight historical 4.0.1 Claude run records remain incomplete. Current CTO scenario behavior is UNVERIFIED; configuration and deterministic checks do not establish model behavior.