agentsclimarketplace

Cc whatsnew

Skill iroha924/mumei/.claude/skills/cc-whatsnew

Use when a new Claude Code version is published, when checking the changelog, or when asked "what was added recently?", "what's new in v2.1.x?", or "fetch the official new features". Reports from the official release notes / changelog in two views - an exhaustive feature summary and an applicability verdict for mumei. Scope via argument (`latest` / `vX.Y.Z` / `vA..vB` / last N); with no argument, choose via AskUserQuestion. After reporting, proposes creating a GitHub issue only when there are mumei adoption candidates.From its SKILL.md

Install
npx -y skills add iroha924/mumei --skill cc-whatsnew

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

6.5 KB, as published. Nobody here has run it

<!-- Role: monitor the official Claude Code changelog and evaluate applicability to mumei Input: version / range (argument or AskUserQuestion) Output: full feature summary + mumei applicability table + (if candidates exist) issue creation Principle: primary sources only (GitHub releases / official docs). Never present memory or search-result summaries as fact. -->

cc-whatsnew

Fetches new Claude Code features from official sources and judges their applicability to mumei in one pass.

Step 1. Determine the scope

Interpret $ARGUMENTS with these rules:

Input exampleInterpretation
latest or emptyrelease notes for the latest version
v2.1.139 / 2.1.139single version
v2.1.135..v2.1.140range (both ends inclusive)
last 5 / last 10cumulative diff of the last N versions

If $ARGUMENTS is empty, choose the scope via AskUserQuestion. Question and options:

  • question: Which range of the Claude Code changelog should be fetched?
  • header: Range
  • options:
    1. Latest version — release notes for the most recent release
    2. Last 5 versions — cumulative diff
    3. Specific version — enter vX.Y.Z via Other
    4. Specify a range — enter vA..vB via Other

For "Specific version" and "Specify a range", use the Other input as-is.

Step 2. Fetch from primary sources

Fetch the original text with WebFetch from one of the following, always. Never report from memory-based guesses or search-result summaries alone:

  • https://github.com/anthropics/claude-code/releases — release page originals
  • https://code.claude.com/docs/en/changelog — official changelog

When a new field or new API appears, additionally fetch for spec confirmation:

  • https://code.claude.com/docs/en/hooks
  • https://code.claude.com/docs/en/skills
  • https://code.claude.com/docs/en/plugins-reference
  • https://code.claude.com/docs/en/commands
  • the feature's dedicated page (e.g. agent-view, model-config)

Label anything that could not be verified as unverified (never present guesses as fact; per the Research discipline section in CLAUDE.md).

Step 3. Organize all features (no mumei filter)

Classify the fetched release notes into the categories below; list each item as a summary that preserves the original meaning, plus the source version.

Categories:

  • Major new features (new commands / screens / workflows)
  • Hook / Plugin
  • MCP
  • Skill / Subagent
  • Settings / Configuration
  • Bug fixes (important ones only)

Compress each item to 1-3 lines; expand deep-dive details (key-binding lists, internals) only in the "Major new features" section.

Step 4. Applicability to mumei

Read the following files before judging:

  • CLAUDE.md
  • merged PR descriptions — where a prior rejection and its reasoning live
  • ARCHITECTURE.md
  • hooks/hooks.json
  • .claude-plugin/plugin.json

Judgment axes per feature:

  1. Does it affect mumei's core (phase gating / Wave commits / review pipeline)?
  2. Has a merged PR already rejected it, and on what grounds?
  3. Would adoption violate KISS / YAGNI?
  4. Forward / backward compatibility risk

Output the verdicts as a table:

| Feature | Verdict | Reason |
|---|---|---|
| name | adopt-candidate / deferred / rejected | 1-2 lines |

Verdict definitions:

  • adopt-candidate — direct benefit to mumei; worth starting implementation
  • deferred — beneficial but premature (official docs not yet updated / forward-compat uncertain, etc.)
  • rejected — conflicts with KISS / YAGNI / design decisions, or irrelevant

Step 5. Propose issue creation

Only when there is at least one adopt-candidate or deferred item, ask via AskUserQuestion:

  • question: Some features look applicable to mumei. Create a GitHub issue?
  • header: Issue
  • options:
    1. Yes, create the issue — create an issue summarizing the candidates via gh issue create
    2. No, revisit later — end with the report only

If there are zero candidates, skip the proposal, note that in one line, and finish.

Step 6. Create the issue (if the user chose yes)

Create an issue with this structure via gh issue create:

gh issue create \
  --repo iroha924/mumei \
  --title "feat: explore Claude Code <range> features for mumei" \
  --label "enhancement,research" \
  --body-file <(cat <<'EOF'
## Summary

Collects the mumei adoption candidates among the features added in Claude Code <range>.

## Candidates

### 1. <feature name>

- Official: <URL>
- Overview: <1-2 lines>
- Verdict: adopt-candidate / deferred
- Considerations:
  - <impact area>
  - <KISS / compatibility risks>

### 2. ...

## Sources

- <release URL>
- <docs URL>
EOF
)

Labels other than enhancement may not exist; check beforehand with gh label list, or fall back to creating with --label enhancement only.

After creation, output the gh issue view <number> URL in one line and finish.

Output format (shared by Steps 3-5)

# What's new in Claude Code <range>

## Major new features
...

## Hook / Plugin
...

## MCP
...

## Skill / Subagent
...

## Settings / Configuration
...

## Bug fixes (excerpt)
...

## Applicability to mumei

| Feature | Verdict | Reason |
|---|---|---|

## Sources

- [...](URL)

What this skill does not do

  • Never presents features unverified against official sources as fact (always use the guess / unverified labels)
  • Never edits mumei code, hooks.json, or docs (stops at issue creation)
  • Never picks a range unilaterally without AskUserQuestion (default to latest only when there is neither an argument nor an AskUserQuestion answer)
  • Never fetches versions outside the requested range (wasted tokens)
  • Never runs git operations other than gh issue create (no commit / push / PR)

Keep looking

Skills are one crate of 627,801. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.