prepare-pr
Skill18 starsChanged 24 days agoApache-2.0View on GitHub
Take a verified Talon change from working tree to a reviewable pull request handoff — diff hygiene, docs set, conventional commits, PR body, CI watch — without pushing or mutating GitHub unless the user asks.
- Commits and pushes
12 other projects use a version of this.
Adopted by
Most recently taken up by
- WikiEducationFoundation/WikiEduDashboardtheir own version, first seen Oct 9, 2026
- TermiX-official/cryptoclawtheir own version, first seen Oct 8, 2026
- mengxi-ream/read-frogtheir own version, first seen Oct 8, 2026
- chattymin/PokeTokenBartheir own version, first seen Oct 5, 2026
- atsushi-green/ds-ai-coding-skillstheir own version, first seen Sep 25, 2026
- NVIDIA/NeMo-Fabrictheir own version, first seen Sep 16, 2026
Used with
Projects with it also use review-pr.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
DockerSkillreview-pr
Review one or more incoming Docker documentation pull requests as a maintainer. Independently validate technical claims, assess editorial fit and information architecture, choose a verdict, and draft exact inline or PR-wide feedback behind a confirmation gate. Use for requests such as "review PR 123", "is this PR correct?", "does this information belong here?", "validate this PR", or "help review backlog PRs". Do not use to maintain or fix a PR you own; use maintain-pr for that.
Also in 3 projects that use this
Other versions12 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 12
- mengxi-ream/read-frog★ 10ksame name
- chattymin/PokeTokenBar★ 557same name
- WikiEducationFoundation/WikiEduDashboard★ 431same name
- NVIDIA/NeMo-Relay★ 246same name
- TermiX-official/cryptoclaw★ 100same name
- atsushi-green/ds-ai-coding-skills★ 86same name
- NVIDIA/NeMo-Fabric★ 75same name
- neul-labs/openclawMU★ 33same name
- pnnl/adept-agentic★ 2same name
- andreilupu/my-claude-skills★ 1same name
2 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository7 other files it gives its agents.
- AGENTS.mdAGENTS.md
- architecture-reviewSkill
- cut-releaseSkill
- implement-issueSkill
- investigate-bugSkill
- respond-to-reviewSkill
- verify-changeSkill
The file
We split this skill into the rules it gives an agent
6 rules. 0 are common, 2 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 11 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
never claim compliance outcomes.
- Very specific: line 19
Run /verify-change and keep the results for the body.
- Very specific: line 25
No AI Co-Authored-By trailers (repository policy in .claude/rules/github-workflow.md).
- Very specific: line 30
Write the body to a file outside the tree ($TMPDIR/pr-<n>.md) or an untracked root file that is never staged.
- Very specific: line 46
Commit and push the feature branch yourself (git push -u origin <branch>), per .claude/rules/github-workflow.md.
- Distinctive: line 52
If the environment has no git credentials, report that and hand over the push command too.
“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Prepare a pull request for $ARGUMENTS.
Before committing
- Be on a feature branch (
feat/…,fix/…,docs/…,chore/…), nevermain. git statusand the complete diff againstmain. Remove anything that is not this change: debug output, scratch files, personal or device files (.bashrc,.idea,.vscode,.claude/settings*, PR body drafts). Stage with explicit paths:git add -- <files>.git diff --check; confirm no credentials, tokens or local paths were added.- Mirrored artifacts move together:
pricing/models.yamlandinternal/pricing/default_models.yaml; a doc table pinned by a test (error contract, governance registry) and its test list; a smoke section and its registration lines (see.claude/rules/contract-surfaces.md). - Docs set for a behaviour change:
CHANGELOG.mdunder## [Unreleased], meeting the quality bar: problem solved, who should care, how to verify (command or test), upgrade or migration impact, one share artifact when applicable;LIMITATIONS.mdwhen the guarantee boundary moved, with a "Backed by" list naming real tests;docs/reference/*for every contract surface touched;examples/*/README.mdwhen a demo path changed;- never claim compliance outcomes; say tamper-evident, not immutable; attribution, not authentication.
- Run
/verify-changeand keep the results for the body.
Commits
- Conventional commits:
type(scope): subjectwith typesfeat fix docs test chore refactor perf; one coherent commit per concern. - The body states why, what was proven, and anything a reviewer must know (migration, follow-up).
- No AI
Co-Authored-Bytrailers (repository policy in.claude/rules/github-workflow.md). - Identity comes from the user's git config; do not invent one.
PR body
Write the body to a file outside the tree ($TMPDIR/pr-<n>.md) or an untracked root file that is never staged. Sections, in order:
- Summary — what is wrong today, proven how (probe test, log, CI run), and the verdict per suspected gap when there are several.
- Decision — the contract chosen and the alternatives rejected, with the product boundary it follows.
- Changes by commit — one line per commit; note compatibility-sensitive surfaces explicitly (schema, evidence, YAML, HTTP/JSON, machine codes, CLI output, exported Go API).
- Regression tests — names, and the statement that each fails without its fix when that was checked.
- Verification matrix — layer, exact command, result; Docker-only rows say "PR checks".
- CI history on this PR — once there is any: commit, failing step, root cause, fixing commit. Never describe a failure as flaky without a mechanism.
- Limitations — what still is not guaranteed, in the same words as
LIMITATIONS.md. - Migration implications — new codes, changed defaults, config to add, re-verification impact.
- Not changed — adjacent things deliberately left alone.
Apply with gh pr create --draft --body-file <path> or gh pr edit <n> --body-file <path>, then confirm it landed: gh pr view <n> --json body -q .body | head. A skipped edit leaves the first draft on the PR indefinitely.
Push yourself, gate the rest
Commit and push the feature branch yourself (git push -u origin <branch>), per .claude/rules/github-workflow.md. Creating or editing the PR, commenting, and merging happen only when the user asks for that action; otherwise end the handoff with the exact command, for example:
gh pr create --draft --title "<type(scope): subject>" --body-file <path>If the environment has no git credentials, report that and hand over the push command too.
After the push
- Watch
gh pr checks <n>(or the checks API). The jobs that gate merge are CI (test, proof-gates, product-demo-smoke, lint, shellcheck, build, docker), Docs, Security, CodeQL, Shortlist Demo and OpenShell Live (live v0.1.3 and the showcase job). - Reproduce every red check with
/verify-changeagainst the commit CI actually ran before touching code. - The PR stays a draft until all checks are green; the user marks it ready and merges.
- Merges on this repository are rebase merges: commit hashes change on
main, and a tag placed on a branch commit will not be an ancestor ofmain. Tag releases onmainafter the merge (/cut-release).
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
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.

