From ba0237a9a0084877ff1f6a59516f7e9fc78e0928 Mon Sep 17 00:00:00 2001 From: "Jason Lee (Titan Zero)" Date: Sun, 4 Oct 2026 15:07:01 +1100 Subject: [PATCH] docs: label policy evidence provenance --- docs/ENGINE_IMPLEMENTATION_AUDIT.md | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/docs/ENGINE_IMPLEMENTATION_AUDIT.md b/docs/ENGINE_IMPLEMENTATION_AUDIT.md index 34002d9..50b5fcd 100644 --- a/docs/ENGINE_IMPLEMENTATION_AUDIT.md +++ b/docs/ENGINE_IMPLEMENTATION_AUDIT.md @@ -40,6 +40,17 @@ php bin/verify.php This source check does not prove database connectivity, queue delivery, host integration, model quality, load behaviour, or production readiness. Catalogue readiness is a separate boundary: `reports/workcore-compatibility.json` records 38 template definitions, 29 ready and nine draft, with zero connected-host-verified templates. +## Policy evaluation and legal provenance + +The 31-case policy result currently linked from the README is historical evidence, not a rerun of the current default branch: + +- The committed result names evaluator commit [`bdbe73c`](https://github.com/Masterleeaus/Interaction-engine/commit/bdbe73cc445b900d29126e081bd5f2341e98ee6f), PHP 8.2.34, seed `20261004`, scenario digest `a1e863c857aa05e70f720224c3f1bb0065f2f1c699fe27e438ce4a64a58ff340`, and 31 cases. +- Current `main` is [`42f7295`](https://github.com/Masterleeaus/Interaction-engine/commit/42f7295219bee40c36717944cf767545024b850d). A source comparison from the evaluated commit to this head shows five follow-up commits, but no changes under `src/Policy/`, `src/Authority/`, `src/AI/`, `scripts/authority-policy-eval.php`, or `evaluations/authority-policy/scenarios.json`. This supports reuse as historical policy evidence; it does not turn the report into a current-head execution record. +- The historical report passed in [CI run 37171514412](https://github.com/Masterleeaus/Interaction-engine/actions/runs/37171514412). The current-head package verification in [CI run 37175375968](https://github.com/Masterleeaus/Interaction-engine/actions/runs/37175375968) did not invoke the policy evaluator. +- Reproduce the evaluation from a clean checkout with `php scripts/authority-policy-eval.php`; after active PR #6 is resolved, rerun it and replace the report only with the resulting current-head commit and CI link. + +Legal provenance is also unresolved on current `main`: the root `LICENSE` path is absent, and GitHub reports no repository license. The README's proprietary-license statement remains the only checked-in guidance. This audit does not invent an MIT or other license; PR #6 proposes a license change and should be resolved by its owner before any legal claim is published. Preserve existing attribution and history until that choice is merged. + ## Recommended next work Implement each no-op method only with its owning contract's intended semantics and a focused test. Until then, keep the methods visible as compatibility scaffolds and keep portfolio language at the deterministic baseline level recorded in [AI_ENGINEERING.md](AI_ENGINEERING.md).