agentleFS
diriendirien/yet-another-agent-harness/.claude/skills/pull-request-msg/SKILL.md

pull-request-msg-with-gh

Skill19 starsChanged 8 months agoMITView on GitHub

Generate a PR_MESSAGE.md file from session context using GitHub CLI. Detects related issues via branch-keyword search, writes a structured PR description with commit subject, summary, test plan, and changelog. Validates with commitlint and markdownlint. Use when preparing a pull request on GitHub.

  • Installs packages

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 1 is distinctive, and 6 are 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.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If gh is not installed, stop and report:

    line 26
  2. Very specific:

    After install, run gh auth login to authenticate

    line 30
  3. Very specific:

    Do not proceed without a working, authenticated gh CLI.

    line 32
  4. Very specific:

    If the user indicates there is no associated issue, omit the Closes #<issue-number> footer entirely.

    line 41
  5. Very specific:

    If the commitlint and markdownlint skills are available in the current context, use them to validate PR_MESSAGE.md.

    line 97
  6. Very specific:

    If neither the skills nor npx are available, fail hard:

    line 110
  7. Distinctive:

    Do not silently skip validation.

    line 116

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
<!-- Copyright 2025-2026 Richard Shade. Licensed under Apache-2.0. --><!-- SPDX-License-Identifier: Apache-2.0 -->

Pull Request Message Generator

Generate a PR_MESSAGE.md file summarizing the current session's work. Requires gh CLI authenticated with the target repository.

Prerequisite check

gh --version 2>/dev/null && gh auth status 2>/dev/null

If gh is not installed, stop and report:

  • Install from https://cli.github.com
  • gh cannot be installed via npm
  • After install, run gh auth login to authenticate

Do not proceed without a working, authenticated gh CLI.

Workflow

  1. Analyze session context — review changes made in the current session. Explore changed files with git diff and git status to understand the full scope of work, in addition to reviewing conversation history. Both sources inform the PR message.
  2. Detect issue number — use the issue detection algorithm below. If the user indicates there is no associated issue, omit the Closes #<issue-number> footer entirely.
  3. Determine commit type — based on the nature of changes (see references/pr-message-format.md for the list of types)
  4. Write PR_MESSAGE.md — using the format in references/pr-message-format.md
  5. Add to .gitignore — add PR_MESSAGE.md if not already present
  6. Validate — run commitlint and markdownlint checks

Issue number detection

The issue number MUST be determined by searching GitHub, NOT by extracting numbers from branch names.

Algorithm

# Step 1: Get branch name
BRANCH=$(git branch --show-current)

# Step 2: Extract keywords (strip leading numbers and hyphens)
# Example: "125-greenops-equivalencies" → "greenops equivalencies"
KEYWORDS=$(echo "$BRANCH" | sed 's/^[0-9]*-//' | tr '-' ' ')

# Step 3: Search GitHub issues
gh issue list --search "$KEYWORDS" --state all \
  --json number,title --limit 5

Evaluation rules

  1. Exactly 1 result with title containing most keywords → use it
  2. 0 results → ask the user for the issue number
  3. 2+ results → show options and ask the user to pick
  4. Ambiguous match → ask the user to confirm

Never extract issue numbers from

  • Branch name digits (e.g., 125- is NOT issue #125)
  • Spec folder names (e.g., specs/125-xyz/)
  • Worktree folder names

These are internal identifiers, not GitHub issue numbers.

Critical rules

  1. No redundancy — summary must add context beyond the commit subject
  2. No ## Commit Message section — the first line IS the commit message
  3. Never modify .markdownlint.json — use CLI flags instead
  4. Trailing newline — file must end with a single newline
  5. Ask when ambiguous — if issue search returns 0, 2+, or unclear results, ask the user

Validation

If the commitlint and markdownlint skills are available in the current context, use them to validate PR_MESSAGE.md. Otherwise, run directly:

# Validate commit message format
cat PR_MESSAGE.md | npx commitlint

# Validate markdown (MD041 disabled — first line is commit subject,
# not heading)
npx markdownlint --disable MD041 -- PR_MESSAGE.md

If neither the skills nor npx are available, fail hard:

  • Install commitlint: npm install -g @commitlint/cli @commitlint/config-conventional
  • Install markdownlint: npm install -g markdownlint-cli

Do not silently skip validation.

Fix any issues and re-validate before reporting success.

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.