Use when cutting a release, writing a CHANGELOG entry, or summarizing what changed since a tag — when the repo's commits follow (or mostly follow) Conventional Commits and a mechanically-accurate section breakdown beats a model's reading of the diff.
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
Parses git log against the Conventional Commits spec and prints a categorized changelog: Features, Fixes, Performance, Refactoring, Docs, Tests, Build, CI, Chore, Other, plus a BREAKING CHANGES section pulled from ! markers and BREAKING CHANGE: footers. Deterministic — the categorization is a fact about the commit message, not a judgment call, so a script produces it more reliably than a model re-reading commit subjects.
Core principle: A changelog section is either correct (matches the commit's own declared type) or it silently drops/miscategorizes a change. Parsing is not the place for approximation.
bash ${CLAUDE_SKILL_DIR}/scripts/generate.sh [from-ref] [to-ref]
from-ref defaults to the most recent tag, or the repo's root commit if there are no tags.to-ref defaults to HEAD.CHANGELOG.md.bash ${CLAUDE_SKILL_DIR}/scripts/generate.sh v1.2.0 HEAD > /tmp/new-entries.md
bash ${CLAUDE_SKILL_DIR}/scripts/generate.sh # since the last tag, to HEAD
Reliable: type(scope)!: subject parsing, breaking-change detection (! or a BREAKING CHANGE: footer), section grouping, short-hash linking, merge commits excluded automatically.
Falls back to "Other": any commit whose subject doesn't start with a known Conventional Commits type (feat/fix/perf/refactor/docs/test/build/ci/chore) — including free-text subjects like "Initial plan" or "Add X". A repo that doesn't consistently use Conventional Commits will produce a large "Other" section; that's an honest signal about the commit history, not a script bug. See it happen: run this against this library's own pre-2.0 history and most entries land in Other because they predate the Conventional Commits convention adopted later.
Not attempted: rewriting or improving commit subjects, deduplicating near-identical entries, or inferring a type the author didn't declare.
The script produces raw categorized entries — it does not write prose framing, a version header, or a date. Add those yourself (or ask Claude to), and edit any "Other" entries that deserve reclassification before they ship in a release. Don't paste the raw output into CHANGELOG.md as the final section without a pass over "Other" and "BREAKING CHANGES" — those two sections are exactly where a human or model read adds real value on top of the deterministic parse.
Categorization is deterministic; the narrative framing around it (a version header, a "why" sentence, what to call out) is not. Do the first with the script, the second with judgment.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Use when evaluating, comparing, installing, or recommending a Claude Code plugin, Agent Skill, MCP server, agent toolkit, or GitHub-hosted AI extension, especially when popularity is weak evidence or the candidate requests shell, network, filesystem, or credential access.
The English summary is being prepared. Showing the original description.
Use before committing, opening a PR, or reporting a task done — when you want ground-truth confirmation there are no conflict markers, live-looking credentials, debug leftovers, untracked TODOs, or oversized files, without spending a model turn re-reading every file yourself.
The English summary is being prepared. Showing the original description.
Use before publishing documentation, merging a docs PR, or reporting a Markdown-heavy repo's navigation as complete — when you want ground-truth confirmation that internal links, heading anchors, and directory links resolve, without opening every file yourself.
The English summary is being prepared. Showing the original description.
Use when the user names a coding, Claude Code, research, documentation, or technology-selection task; asks which prompt to use; says to use the right prompt; or when an agent enters this repository and needs to orient without reading every file.
The English summary is being prepared. Showing the original description.
Use when the user names a coding or Claude Code task ("add OAuth", "set up hooks", "review this PR", "pick a model"), asks which prompt to use, says to "use the right prompt", or when you enter this repo and need to orient without reading every file.
The English summary is being prepared. Showing the original description.
Use when a skill isn't triggering as expected, before adding a new skill (to check for a name collision), during periodic cleanup of an accumulated skills directory, or when asked to review skill quality — checks frontmatter validity, description quality against Skill Discovery Optimization conventions, body size, and cross-scope name collisions.
The English summary is being prepared. Showing the original description.