writing-pull-requests
Skill2 starsChanged 3 months agoMITView on GitHub
Generate conventional-commit PR titles, commit messages, and concise PR descriptions from staged changes or a branch diff. Use when the user asks to create a PR, write a PR title, generate a commit message, or draft a PR description. Output goes directly to the chat — never to a file.
- Commits and pushes
1 other project uses a version of this.
Adopted by
Most recently taken up by
- theexperiencecompany/gaiatheir own version, first seen Nov 13, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- theexperiencecompany/gaia★ 327same name
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 repository19 other files it gives its agents.
- .claude/CLAUDE.mdCLAUDE.md
- api-designSkill
- building-frontend-uiSkill
- debugging-agent-runsSkill
- developing-featuresSkill
- developing-features-tddSkill
- documenting-featuresSkill
- ensuring-cross-platformSkill
- evaluating-ai-outputSkill
- exploratory-testingSkill
- fixing-bugsSkill
- playwright-cliSkill
- reviewing-codeSkill
- running-agent-loopsSkill
- security-reviewSkill
- threat-modelingSkill
- validating-in-realitySkill
- writing-prdSkill
- writing-testsSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 15
Skip / push back: when:
- Very specific: line 24
If learnings/ contains only README.md, proceed with the defaults below.
- Very specific: line 45
Capture a learning (final step, mandatory ask): Before closing the task, ask yourself: *did I encounter a convention, constraint, or trap …
- Very specific: line 81
Never: write a .md file or any other file containing the PR output.
- Distinctive: line 82
Never: run git commit, git add, git push, or any state-changing git command.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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.
Files
Writing Pull Requests
Produce a PR title, a commit message, and a PR description from the current diff, using a conventional-commit format. Output is always printed in chat, never written to a file.
When to apply
Apply when the user says any of: "create a PR", "PR title", "PR description", "commit message", "title for the commit", "write the PR".
Skip / push back when:
- The diff is empty or untracked-only — ask the user what to compare against.
- The diff bundles multiple unrelated changes — propose splitting first; a PR whose title needs "and" is a smell.
- The user already pasted a title/description — don't rewrite unless asked.
Read first (always)
Before doing anything else, list learnings/ and read every file whose name looks relevant to the current PR. Project-specific scopes, type aliases, vocabulary, or hard constraints live there and override the defaults in this SKILL.md. If a learning conflicts with this file, the learning wins — mention it to the user so they can refine it.
If learnings/ contains only README.md, proceed with the defaults below.
Workflow
- Inspect the diff →
git status+git diff --stat <base>..HEAD(or staged with--cached). Identify files added/modified/deleted and the dominant area. → verify: you can name the affected module(s) and the dominant intent (feat/fix/refactor/etc.). - Pick the type from the table below. If two types tie, the change is mixed — flag it and ask whether to split the PR. → verify: one type, defensible from the diff alone.
- Pick the scope from folder names, file paths, or the imports that changed most. Single token, lowercase. Optional — omit if the change is repo-wide. → verify: scope appears in the diff paths.
- Write the summary in imperative mood, ≤72 chars total title length, no trailing punctuation, first letter capitalized after the colon.
→ verify:
<type>(<scope>): <Summary>≤72 chars. - Write the description: one-sentence Objective, ≤5 bullet Changes, optional Notes for edge cases. Whole description ≤150 words. → verify: each bullet starts with a verb; no implementation noise that a reviewer can read from the diff.
- Print the output block (format below) in chat. Do not write a file. Do not run
git commit,git add, orgit push— only the human commits. - Capture a learning (final step, mandatory ask). Before closing the task, ask yourself: did I encounter a convention, constraint, or trap that wasn't in this SKILL.md or in
references/? If yes, append alearnings/YYYY-MM-DD-slug.mdperlearnings/README.md. If no, skip — don't write filler.
Type table
| Type | Use when |
feat | New user-visible capability |
fix | Corrects a defect in existing behavior |
docs | Docs / comments only |
style | Formatting, whitespace, no logic change |
refactor | Restructure, no behavior change |
perf | Performance improvement |
test | Adds or updates tests only |
chore | Build, CI, deps, tooling |
Output format (print verbatim in chat)
TITLE: <type>(<scope>): <Summary>
COMMIT MESSAGE: <type>(<scope>): <Summary>
DESCRIPTION:
## Objective
<one sentence — business outcome, not implementation>
## Changes
- <verb-led bullet>
- <verb-led bullet>
## Notes
<edge cases, limitations, follow-ups — omit the heading if empty>Hard rules
- Never write a
.mdfile or any other file containing the PR output. It goes in the chat. - Never run
git commit,git add,git push, or any state-changing git command. Only the human commits. - Never add AI attribution — no "🤖 Generated with Claude Code" footer, no session links, no AI badges in titles/descriptions/commits, no
Co-Authored-Bytrailer. Output must read as authored by the human. - Never produce a title needing "and" — that means two PRs.
- Never invent scope. If the diff doesn't justify a scope, omit it.
See also
../documenting-features/references/communication.md— answer-first structure (Minto Pyramid); first line = what it does and why it's safe to merge.references/examples.md— wrong-vs-right titles and descriptions, distilled from common LLM mistakes.references/templates.md— full template variants (feature, fix, refactor, revert, breaking change).learnings/— project-specific conventions accumulated over time. Read before generating. Append a new file when you discover a non-obvious convention (seelearnings/README.mdfor the append protocol).
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.

