agentleFS
haojing8312haojing8312/WorkClaw/.agents/skills/workclaw-release-prep/SKILL.md

workclaw-release-prep

Skill141 starsChanged 7 months agoApache-2.0View on GitHub

Use when preparing a WorkClaw desktop release and you need AI to recommend the next version number, summarize changes since the last tag, and draft bilingual Chinese plus English release notes for human confirmation before publishing.

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 7 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

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

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

    Use this skill before cutting a WorkClaw release.

    line 9
  2. Very specific:

    Do not use this skill to push commits, create tags, or publish releases.

    line 17
  3. Very specific:

    Before drafting release notes, build a short candidate user-visible changes list from:

    line 31
  4. Distinctive:

    commit titles and commit bodies

    line 32
  5. Distinctive:

    test names that describe user-facing behavior

    line 33
  6. Distinctive:

    Default rule: if a commit message or test name names a concrete model, provider, or desktop behavior, treat it as a release-note candidate …

    line 37
  7. Distinctive:

    Keep release notes concise and user-facing

    line 80
  8. Distinctive:

    Avoid speculative claims not backed by repo changes

    line 83
  9. Distinctive:

    Start from the candidate highlight list, then compress.

    line 85
  10. Distinctive:

    When a change improves compatibility for a named provider or model, prefer stating that user-facing outcome plainly

    line 86
  11. Very specific:

    If a candidate highlight is omitted from the final notes, explain why in Intentionally omitted

    line 87

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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

WorkClaw Release Prep

Overview

Use this skill before cutting a WorkClaw release. It recommends the next version, explains why, and drafts bilingual release notes. It does not publish anything.

When to Use

  • The user wants a release recommendation
  • The user wants AI to decide between patch, minor, or major
  • The user wants bilingual Chinese + English release notes
  • The user wants a confirmation-ready release package before tagging

Do not use this skill to push commits, create tags, or publish releases. Use workclaw-release-publish after confirmation.

Inspect First

  • git tag --sort=-creatordate
  • git log <last-tag>..HEAD --oneline
  • git log <last-tag>..HEAD --format=%H%n%s%n%b%n==END==
  • git diff --name-only <last-tag>..HEAD
  • For high-signal commits, inspect git show --stat --summary <sha>
  • apps/runtime/package.json
  • apps/runtime/src-tauri/Cargo.toml
  • apps/runtime/src-tauri/tauri.conf.json
  • .github/release-windows-notes.md
  • Any release-impacting docs or plans the user wants included

Before drafting release notes, build a short candidate user-visible changes list from:

  • commit titles and commit bodies
  • test names that describe user-facing behavior
  • provider or model names mentioned explicitly, such as Qwen, DeepSeek, OpenAI, Claude
  • compatibility, routing, transport, recovery, installer, branding, and skill/runtime keywords

Default rule: if a commit message or test name names a concrete model, provider, or desktop behavior, treat it as a release-note candidate unless there is strong evidence it is purely internal.

Version Recommendation Rules

  • Recommend patch for bug fixes, stability hardening, recovery improvements, small UX polish, or non-breaking desktop flow fixes
  • Recommend minor for clear new user-facing capabilities, new workflows, new navigation models, or meaningful product surface expansion
  • Recommend major only for breaking changes, migrations, compatibility breaks, or strong user behavior changes that require explicit upgrade communication

When uncertain, provide:

  • one recommended version
  • one conservative alternative
  • a short rationale for both

Required Output

Use this shape:

## Release Prep
- Last tag:
- Recommended version:
- Alternative version:
- Why:
- Release scope:
- Files or areas reviewed:

## Candidate Highlights
- Included in notes:
- Intentionally omitted:
- Follow-up checks:

## Release Notes Draft
### 中文
- ...

### English
- ...

## Confirmation Needed
- Confirm version:
- Confirm release notes:
- Ready for publish skill: yes | no

Drafting Rules

  • Keep release notes concise and user-facing
  • Group changes by outcomes, not by files
  • Mention desktop installer guidance only if relevant
  • Avoid speculative claims not backed by repo changes
  • Chinese and English sections should say the same thing, not different things
  • Start from the candidate highlight list, then compress; do not draft directly from file paths alone
  • When a change improves compatibility for a named provider or model, prefer stating that user-facing outcome plainly
  • If a candidate highlight is omitted from the final notes, explain why in Intentionally omitted

Common Mistakes

  • Recommending minor for pure bugfix bundles without a new capability
  • Recommending patch when a clearly new workflow shipped
  • Mixing internal implementation detail into user-facing release notes
  • Reading only changed file paths and missing user-visible gains described in commit bodies or tests
  • Treating provider/model compatibility fixes as internal-only by default when they change whether users can actually run that provider
  • Proceeding to publish without explicit human confirmation

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.

Reviews can't be read right now.

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.