What does this skill do, and when should you use it?
This skill automatically scans merged PRs and commits since a given reference, extracting titles, labels, types, and signals. It categorizes changes into features, bug fixes, documentation, dependencies, breaking changes, and security. The output is a categorized draft release note saved to RELEASE_NOTES_DRAFT.md, plus a state file update. Designed to be draft-only (L1 level), it will not create PRs, releases, or tags without explicit human approval.
Runs git log --merges --oneline <last-tag>..HEAD to fetch merged commits. Collects PR labels, milestones, and commit messages. Categorizes changes into predefined categories. Writes RELEASE_NOTES_DRAFT.md with the categorized draft, and updates changelog-drafter-state.md with the scan window. Ensures breaking changes and security items are surfaced explicitly.
- A maintainer wants a quick summary of recent merged changes before a release.
- A developer needs to automatically build a draft release note based on git history since the last tag.
- A team wants release notes organized by category (features, bug fixes, etc.) to streamline release announcements.
- A maintainer prepares a draft for a weekly release and tracks the scan window for review.
How do you install this skill?
- No guidance on handling sensitive info (e.g., secrets in PRs) which could leak into drafts.
- No error handling; failures in git commands or missing tags lack clear diagnostics.
- No installation/dependency notes, may hinder actual use.
- Lacks versioning, changelog, and maintenance responsibility disclosure, affecting long-term maintainability.
- Repo may depend on overseas services, but skill has no network dependency; mainland accessibility unverified.
- Shell / CLI
- Local filesystem
Clone or copy the repository into your environment and place the skill folder (starters/changelog-drafter-opencode/skills/changelog-scan/) in your agent's skills directory. For this mono-repo, refer to the README for installation of the whole collection; no separate installation instructions are provided for this specific skill.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/cobusgreyling/loop-engineering.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/starters/changelog-drafter-opencode/skills/changelog-scan" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- scan changelog from v1.2.0
Invoke the skill in your agent, specifying the last tag (e.g., 'scan changelog from v1.2.0'). The skill will run git commands and produce a draft. The draft will be in RELEASE_NOTES_DRAFT.md; review and refine manually. It will not tag or create PRs without explicit approval.
What are this skill's strengths and limitations?
- Automates extraction of merged PRs and commits, reducing manual effort.
- Provides structured output with standard categories, making it easy to organize.
- Designed as a safe L1 draft, it never publishes or tags automatically.
- Updates a state file to track the scan window, aiding review.
- No specifics on how PR labels or milestones are accessed (only described, not commands).
- No documented test suite or validated agent platform (except OpenCode as an example).
- The example is for OpenCode; other agents may need adjustments.
- Unclear how to handle repositories without labels or milestones; relies on commit messages which may be inconsistent.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Changelog Scan Skill this page | 45 · Not recommended | ★ 11k | 3d ago | MIT |
| Changelog Scanner | 43 · Not recommended | ★ 11k | 3d ago | MIT |
| Wiki Changelog Assistant ✓ Microsoft · Official | 51 · Use with care | ★ 2.7k | 3mo ago | MIT |
| DeepChat Release Assistant | 46 · Use with care | ★ 6.4k | 3d ago | Apache-2.0 |
| Release Notes Drafting Skill | 32 · Not recommended | ★ 11k | 3d ago | MIT |
How did FollowSkills review this skill?
The skill only reads git history and produces a draft, with no network exfiltration or file writes, exercising least privilege; clearly requires explicit human approval before publishing or tagging, aligning with safety. However, it does not mention sensitive-data handling, dependency security, or rollback, and data-flow transparency is limited. Repo has SECURITY.md but skill does not reference it.
Steps are mostly self-consistent with concrete git commands, but no error handling or abnormal-input feedback; on failure no clear explanation. Depends on git but no install/dependency notes in skill. Static review cannot verify execution, capped below static maximum.
Targets changelog drafting clearly, but lacks declared non-fit boundaries or edge cases; user-invocable but precise trigger conditions not specified. Environment fit unknown, no Chinese support or overseas dependency noted, mainland availability unverified.
Clear front matter and categories, structured; but lacks install/dependency notes, examples, FAQ, known limitations, versioning, changelog, and maintenance responsibility. MIT license clear but not referenced in skill file.
Produces categorized draft and updates state file, but no output format spec or example, cannot confirm direct usability. Comparative benefit vs manual not stated; static cannot verify actual results.
Only SKILL.md and repo-level CI; no tests specific to this skill or third-party execution evidence. Repo has CI workflows but not covering this skill. Static assessment limits evidence.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →