agentleFS
carlkiblercarlkibler/agent-skills/skills/changelog-writer/SKILL.md

changelog-writer

Skill6 starsChanged 4 months agoMITView on GitHub

Draft plain-language changelog entries from Git history, optionally updating CHANGELOG.md.

21 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use regex-builder, test-writer and 1 more.

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.
  • jMertaSkill

    regex-builder

    Use when the user asks to create, debug, explain, validate, or optimize a regular expression or replacement for a specific engine. Verifies representative matches, non-matches, captures, escaping, portability limits, and backtracking risk in the target runtime.

    Also in 3 projects that use this

  • leonardomsoSkill

    test-writer

    Generate comprehensive Vitest tests for code examples in JavaScript concept documentation pages, following project conventions and referencing source lines

    Also in 5 projects that use this

  • glinckerSkill

    readme-generator

    Generates comprehensive README.md files for software projects by analyzing codebase structure

    Also in 6 projects that use this

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 10 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 19 edited versions we read keep these.

  • 10

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before a release, version bump, or deployment

    line 19
  2. Distinctive:

    When a CHANGELOG.md needs updating

    line 20
  3. Distinctive:

    When you need release notes for a blog post, email, or app store update

    line 21
  4. Distinctive:

    When commits describe implementation but not user impact

    line 22
  5. Distinctive:

    When the user wants a changelog from scratch with no baseline: ask for a starting commit or date instead of guessing

    line 27
  6. Distinctive:

    Write in plain English: "You can now do X" / "Fixed: X no longer happens"

    line 68
  7. Distinctive:

    NEVER include: commit SHA, file names, function names, internal terminology

    line 69
  8. Distinctive:

    No filler: Don't write "Various improvements and bug fixes" — if you can't describe it specifically, skip it.

    line 98
  9. Distinctive:

    Read the existing file.

    line 130
  10. Distinctive:

    If any check fails, revise before presenting.

    line 144

“Common” is measured: kept by at least 6 in 10 of the 19 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Changelog Writer

Turns git commits into changelog entries humans want to read. Covers the gap between "git log --oneline" (too technical) and "nothing" (too common).

When to Use

  • Before a release, version bump, or deployment
  • When a CHANGELOG.md needs updating
  • When you need release notes for a blog post, email, or app store update
  • When commits describe implementation but not user impact

When NOT to Use

  • For internal-only changes with no user-visible effect
  • When the user wants a changelog from scratch with no baseline: ask for a starting commit or date instead of guessing

<process>

Phase 1: Determine the Scope

Ask or infer:

  1. Git range — "since last release" (git tags), a specific SHA, or a branch diff. Default: git log $(git describe --tags --abbrev=0)..HEAD if tags exist, else ask for a starting point (a short-history repo breaks HEAD~20). Confirm the resolved range before drafting.
  2. Audiences — which sections to generate. Default: all three (Users, Developers, Operators). Skip any that aren't relevant.
  3. Format — CHANGELOG.md (Keep a Changelog format), Markdown for a blog post, or plain text for an email.
  4. Version — if known, use it. Otherwise leave as [NEXT VERSION].

Detect git range automatically

# Check if tags exist
git describe --tags --abbrev=0 2>/dev/null && \
  echo "Use: git log $(git describe --tags --abbrev=0)..HEAD --pretty=format:'%H %s' --no-merges" || \
  echo "No tags found — use: git log HEAD~20..HEAD --pretty=format:'%H %s' --no-merges"

Phase 2: Read the Git History

Run:

git log <range> --pretty=format:"%H %s" --no-merges

For each commit whose subject does not make the user impact clear, read the body and diff summary (git show --stat <SHA>). PR-merge or squash commits may need gh pr view for intent. Do not guess impact from a subject alone.

Group commits by type if using conventional commits (feat, fix, docs, chore, refactor, perf, test, build, ci). Otherwise, read the commit message and classify by likely intent.

Phase 3: Classify and Draft

For each commit, decide:

Affects users? → Include in "Users" section

  • New features, UI changes, behavior changes, bug fixes visible to users
  • Write in plain English: "You can now do X" / "Fixed: X no longer happens"
  • NEVER include: commit SHA, file names, function names, internal terminology

Affects developers integrating or extending? → Include in "Developers" section

  • API changes, new hooks/events, SDK changes, breaking changes, new endpoints
  • Can include technical terms, but explain impact: "Breaking: X now returns Y instead of Z"

Affects deployment, configuration, or operations? → Include in "Operators" section

  • New environment variables, config changes, migration steps, infra changes
  • Be specific: "New required env var: FEATURE_X_ENABLED (default: false)"

Internal only? → SKIP (refactors, test changes, CI changes, code style)

Transformation examples

Raw commitTranslated entry
fix: resolve null pointer in UserService.processPaymentFixed a bug that caused payment processing to fail for some users.
feat: add dark mode toggle to settings panelYou can now switch to dark mode in Settings.
perf: replace O(n²) sort in feed rankingFeeds with many items now load significantly faster.
chore: upgrade postgres driver to 16.x(Operators) Updated database driver — no action required unless pinning the old version.
feat: add /v2/webhooks endpoint with retry logic(Developers) New: /v2/webhooks endpoint with automatic retry on failure. See API docs.
refactor: extract PaymentGateway interfaceSKIP — internal only.

Phase 4: Write the Entries

Rules for writing:

  • Users section: Plain English. No jargon. Lead with the user benefit, not the implementation. Max 2 sentences per entry. Use "You can now...", "Fixed...", "Improved..."
  • Developers section: Technical but clear. Always state the migration path for breaking changes.
  • Operators section: Imperative and specific. "Set X to Y before deploying." "Run migration Z."
  • No filler: Don't write "Various improvements and bug fixes" — if you can't describe it specifically, skip it.
  • Group related items: Multiple commits fixing the same area → one entry.

Phase 5: Format Output

Keep a Changelog format (default)

## [NEXT VERSION] - <today's date, YYYY-MM-DD>

### For Users

- **New:** [feature description]
- **Fixed:** [bug fix description]
- **Changed:** [behavior change and why]

### For Developers

- **Breaking:** [what changed, how to migrate]
- **New:** [new API/hook/event]

### For Operators

- **Action required:** [what to do before deploying]
- **New config:** [variable name and purpose]

Blog post / email format

Omit the markdown headers. Write short prose paragraphs instead of bullet lists. Friendly tone, no technical filler.

If the user wants to update CHANGELOG.md

  1. Read the existing file.
  2. Identify the position of the # Changelog header (or top of file).
  3. Prepend the new section immediately after the header, before any existing entries.
  4. Preserve all existing content verbatim.
  5. Show the entry first; write it back only on confirmation. If the file uses a different heading style, match it.

Phase 6: Review Pass

Before presenting, check:

  • Are all user-visible changes covered?
  • Is there any internal jargon in the Users section?
  • Are breaking changes clearly labeled?
  • Would a non-technical user understand the Users section?

If any check fails, revise before presenting.

</process>

<anti_patterns>

Anti-Patterns

Don'tDo instead
Include commit SHAs or branch names in the Users sectionWrite about what the user experiences, not what the developer did
Write "Various improvements and bug fixes"Name the specific fix or improvement, or skip it
Put refactors and test changes in any audience sectionInternal changes have no changelog entry
Use function names, file names, or class names in the Users sectionTranslate to user experience: "payment processing" not "UserService.processPayment"
Write passive vague entries ("Performance has been improved")Name the scenario: "Large dashboards now load in under 2 seconds"
Produce a Developers section entry without migration guidance for breaking changesAlways include "Before: X. After: Y. To migrate: Z."
Produce an Operators section entry without the exact variable name or commandBe specific: REDIS_MAX_CONNECTIONS=50 not "configure Redis appropriately"
Cover 0 commits in a section when none qualifyOmit the section entirely if it has no entries

</anti_patterns>

<success_criteria>

The changelog is complete when:

  • Every user-visible change has an entry in plain English
  • No commit SHAs, file names, or function names appear in the Users section
  • Breaking changes are labeled and include a migration path
  • Operator actions are imperative and specific
  • Internal-only commits (refactors, CI, tests) are excluded
  • Related commits are grouped, not listed individually
  • The user has been offered to write/update CHANGELOG.md if applicable

</success_criteria>

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.