From aa5d712aea07d95a90fd147555f78753f297ed1d Mon Sep 17 00:00:00 2001 From: Serhii Date: Sat, 18 Apr 2026 10:51:36 +0300 Subject: [PATCH] docs(contributing): add AI usage policy section (closes #106 discussion) Surfaces the three-point AI usage policy Jason articulated in #106 by placing it in CONTRIBUTING.md between "Code Standards & Tooling" and "Running the Test Suites", so first-time AI-assisted contributors see the expectation in the onboarding doc rather than having to discover it via issue archaeology. Text is the #106 comment verbatim with two small typo fixes only (use if AI -> use of AI; doesnt -> doesn't); no structural or stylistic rewriting. Per Jason's green light to open this PR in NawfalMotii79/PLFM_RADAR#106 (comment-4273144522). --- CONTRIBUTING.md | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 83e5965..448b518 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -42,6 +42,14 @@ Never run bare package installation commands without the `sfw` prefix. - **System-Level Invariants**: - Whenever adding code, verify that system-level invariants (across module, process, and chip boundaries) hold true. +## AI Usage Policy + +The use of AI is permitted but we have to make sure that the quality and control of the codebase doesn't depend on the agents but the maintainer pushing the changes, meaning they are fully responsible for the code they commit. + +1. **Human Accountability** — The committing engineer is fully responsible for AI-generated code as if they wrote it. Every PR must be understood and defensible by a human. +2. **Mandatory Review** — No raw AI output may be committed unread. AI code must pass the same review bar as hand-written code. +3. **Full CI Before Commit** — All AI-assisted changes must pass the complete CI suite locally (lint, unit, regression, cross-layer) before commit. + ## Running the Test Suites We use GitHub Actions for CI, which runs four main jobs on every PR. Run these locally before pushing.