agentleFS
tenforward-audiotenforward-audio/swe-harness/.agents/skills/integrate-reviewed-change/SKILL.md

integrate-reviewed-change

Skill0 starsChanged 3 months agoMITView on GitHub

Integrate an already reviewed candidate into the coordinating checkout after the user explicitly authorises that integration. Use for a pinned commit or branch whose review evidence and common base are known. Do not use for implementation, unreviewed conflict resolution, inferred acceptance, push, release, branch deletion, or worktree removal.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the root AGENTS.md, the contributing guide, the relevant lifecycle card when one exists, and the completed review evidence.

    line 16
  2. Very specific:

    Verify that a named branch still points to the reviewed commit.

    line 21
  3. Distinctive:

    If the integration method would materially change history or project behavior and neither the user nor project instructions selected it, …

    line 28
  4. Distinctive:

    Use the explicitly selected non-history-rewriting integration method.

    line 43
  5. Distinctive:

    Do not improvise substantive conflict resolutions.

    line 45
  6. Distinctive:

    Run the configured affected-layer checks and the required full gate after the integrated tree is coherent.

    line 53
  7. Distinctive:

    After successful verification, if the user has accepted or otherwise explicitly declared the candidate complete, name the retained local …

    line 59
  8. Very specific:

    Do not infer acceptance or card closure from integration.

    line 66

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Integrate a Reviewed Change

Execution mode

inline

Confirm authority and reviewed identity

  1. Require an explicit user request to integrate or merge. Review completion or acceptance evidence alone is not integration authority.
  2. Read the root AGENTS.md, the contributing guide, the relevant lifecycle card when one exists, and the completed review evidence.
  3. Resolve the exact reviewed candidate commit, reviewed common base, coordinating checkout, target ref, and requested integration method.
  4. Verify that a named branch still points to the reviewed commit. If it advanced, send the new candidate through review-project-change before integration.
  5. Inspect target dirtiness and unrelated user changes. Stop when the integration cannot be isolated without risking them.

If the integration method would materially change history or project behavior and neither the user nor project instructions selected it, ask before mutating the target. Do not infer authority for rebase, history rewriting, or a remote operation.

Integrate only reviewed scope

  • Recheck the candidate diff from its common base and compare it with changes on the target since that base. Do not integrate a known failing or incomplete candidate.
  • Audit root instructions and .agents/ independently. Reject stale queue, board, policy, or evidence copies from a background lane. Apply legitimate reviewed canonical-state changes only in the designated coordinating checkout. Preserve lifecycle-tracking checkpoints already made on the canonical target; they are not required to appear on the pinned candidate branch.
  • Use the explicitly selected non-history-rewriting integration method. Preserve authorship and commit identity where the method allows it.
  • Do not improvise substantive conflict resolutions. A conflict that changes reviewed behavior creates a new candidate that must return to development and review.
  • Integration authority does not include push, tag, publication, release, branch deletion, worktree removal, or cleanup of preserved evidence.

Verify and hand off

  • Run the configured affected-layer checks and the required full gate after the integrated tree is coherent.
  • Inspect the resulting diff and history for intended scope, canonical-state preservation, and unrelated changes.
  • Report the integrated candidate and resulting commit, checks, conflicts or deviations, and observed versus unobserved evidence.
  • After successful verification, if the user has accepted or otherwise explicitly declared the candidate complete, name the retained local candidate worktree and branch and offer to remove them. State that cleanup is optional, requires a separate unambiguous agreement, and will not delete a remote branch or durable review evidence. If the user agrees, use clean-up-worktree. If tracked acceptance is still pending, retain the artifacts without offering cleanup yet.
  • Do not infer acceptance or card closure from integration. Use manage-project-work only when the user explicitly requests the corresponding lifecycle transition.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.