agentleFS
pawellachpawellach/pm-ai-prompts/skills/github-pr-to-release-notes/SKILL.md

github-pr-to-release-notes

Skill0 starsChanged 59 days agoMITView on GitHub

Generate structured release notes from merged GitHub PRs for a given milestone or date range

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    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
  1. Very specific:

    Read merged PRs from a GitHub repository and generate ready-to-publish release notes in all three formats defined in …

    line 8
  2. Distinctive:

    Use this skill when the user says:

    line 12
  3. Distinctive:

    If the user hasn't specified a repo, ask: "Which GitHub repository?

    line 50
  4. Distinctive:

    If a PR has no labels, classify by title keywords:

    line 66
  5. Very specific:

    PRs without good titles: If a significant portion of PRs have commit-style titles ("fix stuff", "WIP", "update"), flag them in the output …

    line 121
  6. Very specific:

    Monorepo with multiple services: Ask the user which subdirectory/package/service to scope to, or whether notes should cover all packages.

    line 125

≈ “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

GitHub PR → Release Notes

Read merged PRs from a GitHub repository and generate ready-to-publish release notes in all three formats defined in prompts/release-notes-writer.md.

When to use this skill

Use this skill when the user says:

  • "Generate release notes for v2.3.0"
  • "What shipped in the last sprint / this week / this milestone?"
  • "Pull the PR list and write the release notes"
  • "Prepare the changelog from GitHub"

Prerequisites

  • gh CLI installed and authenticated (gh auth status)
  • Repository accessible from the current machine

Step 1 — Identify the scope

Determine which PRs to include. Ask the user for one of:

Option A — By milestone:

gh pr list --repo OWNER/REPO --state merged --search "milestone:v1.2.3" --json number,title,labels,mergedAt,body --limit 100

Option B — By date range:

gh pr list --repo OWNER/REPO --state merged --json number,title,labels,mergedAt,body --limit 100 | \
  jq '[.[] | select(.mergedAt > "2026-01-01T00:00:00Z" and .mergedAt < "2026-02-01T00:00:00Z")]'

Option C — By tag comparison (since last release):

# Find the last release tag
gh release list --repo OWNER/REPO --limit 5

# List commits between tags
git log v1.1.0..v1.2.0 --oneline

# Match commits to PRs
gh pr list --repo OWNER/REPO --state merged --search "base:main" --json number,title,labels,mergedAt --limit 100

If the user hasn't specified a repo, ask: "Which GitHub repository? (owner/repo format)"

Step 2 — Categorize PRs by label

Map GitHub labels to release note categories:

GitHub LabelRelease Note Category
breaking-change, breakingBreaking Changes
feature, enhancement, newNew Features
improvement, refactor, perfImprovements
bug, fix, bugfixBug Fixes
security, vulnerability, cveSecurity Updates
deprecation, deprecatedDeprecations
No label / chore, docs, ci(ask user whether to include)

If a PR has no labels, classify by title keywords:

  • Starts with "feat:", "feature:", "add " → New Features
  • Starts with "fix:", "bug:", "patch " → Bug Fixes
  • Starts with "perf:", "optim" → Improvements
  • Starts with "BREAKING", "breaking:" → Breaking Changes

Flag any PRs where classification is ambiguous — do not silently skip them.

Step 3 — Extract PR metadata

For each PR, capture:

  • PR number and title
  • Author
  • Linked issues (from PR body — Fixes #123, Closes #456)
  • Merge date
  • Description / body (first paragraph, cleaned of markdown links and checklists)

Skip PRs labeled: wip, do-not-merge, draft, internal-only — note the count skipped.

Step 4 — Generate release notes

Pass the categorized PR list to the three-format structure from prompts/release-notes-writer.md:

Format A — Technical (for developers):

  • List each PR with its number, title, and brief description
  • Breaking changes section first, always
  • Migration steps if any breaking changes exist

Format B — Product (for users and customer success):

  • Translate PR titles into user-facing benefit language
  • Group by feature area, not by PR
  • Skip internal PRs (infra, CI, chore, refactor unless user-visible)

Format C — Executive (for stakeholders):

  • Maximum 8 bullets total
  • Business impact language only
  • Security patches always mentioned if present

Step 5 — Output

Produce all three formats plus:

## PR Summary
- Total PRs merged: N
- Breaking changes: N
- New features: N
- Bug fixes: N
- Security patches: N
- PRs skipped (WIP/internal): N
- Ambiguous (manual review needed): [list]

Common issues

PRs without good titles: If a significant portion of PRs have commit-style titles ("fix stuff", "WIP", "update"), flag them in the output as [NEEDS REVIEW] rather than generating poor release notes.

Large release (50+ PRs): Group related PRs before writing — don't produce 50 bullet points. Summarize by theme.

Monorepo with multiple services: Ask the user which subdirectory/package/service to scope to, or whether notes should cover all packages.

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.