Skip to content

feat: incorporate the Drupal.org AI contribution policy into bundled skills - #375

Merged
mglaman merged 2 commits into
mainfrom
feat/ai-contribution-policy-skills
Sep 4, 2026
Merged

mglaman merged 2 commits into
mainfrom
feat/ai-contribution-policy-skills

Conversation

@mglaman

@mglaman mglaman commented Sep 4, 2026

Copy link
Copy Markdown
Owner

What changed

The bundled agent skills now carry Drupal.org's policy on the use of AI when contributing as operational checkpoints, so contributors using agents with this CLI adhere to it by default.

Why

Agents driving drupalorg produce code and text the user submits under their own account. The policy holds the human responsible for every line, forbids drive-by contributions, and requires disclosure of significant AI-generated work. Violations lead to account bans. None of that was represented in the skills.

How

  • New shared checklist at skill-data/drupalorg-cli/references/ai-contribution-policy.md, served by skill:get drupalorg-cli --full. It covers: read the thread first, minimal and explainable diffs, no unverified dependencies, green pipeline before hand-off, never push to someone else's MR unannounced, the AI-Generated: Yes (...) disclosure format, summaries in the user's own words, and staying responsive to feedback.
  • The drupalorg-cli skill and the installed discovery stub each gain a short "Contribution policy" section with the non-negotiables.
  • drupalorg-work-on-issue gains [POLICY] checkpoints: fetch comments and report prior attempts in Step 1, stop when the MR author is not the user in Step 3, forbid out-of-scope refactors in the work loop, require a drafted disclosure before flipping to Needs review, and remind the user about follow-up.
  • drupalorg-issue-summary-update refuses updates with no verified insight, hands over a draft to rewrite rather than paste, and adds a disclosure step.
  • Workflow references and the README carry short pointers.

Known gap

--with-comments is silently ignored for GitLab work items, so the skills tell agents to fall back to glab issue view --comments for those. Tracked in #374.

Testing notes

  • vendor/bin/phpunit: 212 tests pass.
  • php drupalorg skill:get drupalorg-cli --full includes the new ai-contribution-policy reference.
  • The CI assertion that the stub still points at drupalorg skill:get drupalorg-cli holds.

🤖 Generated with Claude Code

…skills

Agents driving drupalorg produce code and text the user submits under
Drupal.org's policy on the use of AI when contributing. The skills now
carry that policy as operational checkpoints: read the thread before
writing, keep diffs minimal and explainable, pass local checks and the
MR pipeline before hand-off, never push to someone else's MR unannounced,
draft the required AI-Generated disclosure, and hand summaries over as
drafts in the user's own words.

A shared checklist lives in references/ai-contribution-policy.md and is
served by skill:get drupalorg-cli --full. The discovery stub, main skill,
work-on-issue and issue-summary-update skills, workflow references, and
README point to it.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@mglaman
mglaman merged commit 40c3aec into main Sep 4, 2026
9 checks passed
@mglaman
mglaman deleted the feat/ai-contribution-policy-skills branch September 4, 2026 14:22
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant