agentleFS
akshay-diwadkarakshay-diwadkar/first-four-hours/skills/workbench/pr/SKILL.md

pr

Skill1 starsChanged 4 days agoMITView on GitHub

Write a grounded, reviewer-oriented pull request description for the current branch and open or update its PR only after explicit user confirmation. Use when a finished branch needs a PR or a refreshed description; never merges.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Distinctive:

    Write what a reviewer needs and the diff cannot show, check it mechanically, and open the PR only after the user confirms the exact text.

    line 10
  2. Very specific:

    Read Description guide, then write the draft outside the repository, next to <runs.jsonl>.

    line 27

≈ “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

PR

Write what a reviewer needs and the diff cannot show, check it mechanically, and open the PR only after the user confirms the exact text. Never merge, approve, or edit the branch's code.

Ground

  1. Fix the base ref (for example origin/main) and read the whole git diff <base>...HEAD. If gh pr view <branch> --json state shows OPEN, read Update first.
  2. Find the authority: the linked issue (gh issue view <n>) or a sealed .first-four-hours/blueprints/*/blueprint.md. Link or quote its intent, decisions, and acceptance criteria; never re-author them as a second source of truth. With neither, say so and take intent from the user.
  3. From a clean worktree at the commit you will push, run each verification you will claim through record, one arg per argv element:
python <skill-root>/scripts/cli.py --repo-root <repo> --input runs=<runs.jsonl> \
  --input arg=<argv0> --input arg=<argv1> --format json record

Its result gives the real exit_code and the command string to cite.

Draft

Read Description guide, then write the draft outside the repository, next to <runs.jsonl>.

Check

python <skill-root>/scripts/cli.py --repo-root <repo> --input draft=<draft.md> \
  --input runs=<runs.jsonl> --input base=<base> --format json check

Repair each diagnostic in the draft, or record the missing run, and rerun until it passes. The script checks only grounding, recorded runs, and Mermaid shape, never prose quality; the content and every diagram are your judgment.

Confirm and open

Show the user the title and the complete checked draft, and ask whether to open the PR, and to push the branch first when its upstream is missing or behind. Only an explicit yes authorizes it; silence or an earlier approval does not. After a yes, push if needed, then:

python <skill-root>/scripts/cli.py --repo-root <repo> --input draft=<draft.md> \
  --input runs=<runs.jsonl> --input base=<base> --input title=<title> \
  --input confirm=<body_sha256 from check> --format json publish

confirm binds the exact text the user saw; after any edit, check, show, and ask again. Report the PR URL. End with one optional next step fitting this outcome: a fenced, self-contained prompt naming any repository, issue, PR, or artifact it needs, or say none is needed. Only a "yes" starts it, via any skill it names, but the user runs a charter, diagnose, or ticket step as its /<skill> command. Never merge the PR.

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.