Skip to content
Agentmart
Submit a tool

changelog-review

Use when a Claude Code plugin project needs auditing against live official docs — hooks.json, SKILL.md, plugin.json, MCP config — for misuse, broken hook…

Skill by Tamircohen28

IndexedMIT★ 2
Write a review

Chalk Labs

Need help deploying changelog-review?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/tamircohen28-changelog-review

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Use when a Claude Code plugin project needs auditing against live official docs — hooks.json, SKILL.md, plugin.json, MCP config — for misuse, broken hook paths, stale patterns and missed capabilities. Invoked by repo-standards polish phase 6; also callable by any skill needing a live-doc-grounded answer about Claude Code hooks, skills, plugin.json or MCP. Version agreement across manifests and badges is delegated to plugin-version.json + scripts/check-version-truth.sh. Never invoke directly — use repo-standards.

Use cases: documents, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md and code. Checked Nov 1, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found
Runs as
Shell scripts

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/changelog-review

What's inside

Skill name
changelog-review
Description
Use when a Claude Code plugin project needs auditing against live official docs — hooks.json, SKILL.md, plugin.json, MCP config — for misuse, broken hook paths, stale patterns and missed capabilities. Invoked by repo-standards polish phase 6; also callable by any skill needing a live-doc-grounded answer about Claude Code hooks, skills, plugin.json or MCP. Version agreement across manifests and badges is delegated to plugin-version.json + scripts/check-version-truth.sh. Never invoke directly — use repo-standards.
License
MIT
Metadata
provider: developer-workflow · capability: documentation · updated-date: 2026-08-19
Instructions
362 lines in SKILL.md · view on GitHub

Before installing

  • Ships executable scripts

Agent Skills spec issues

  • compatibility: must be text of at most 500 characters
  • allowed-tools: must be a space-separated string
  • files: referenced but missing: scripts/check-version-truth.sh

Files (6)

  • SKILL.md
  • evals.json · evals/
  • trigger-evals.json · evals/
  • urls.md · references/
  • check-skill-frontmatter.sh · scripts/
  • validate-plugin-json.sh · scripts/

SKILL.md (first lines)

# changelog-review

You are a Claude Code documentation expert. Your answers are grounded **exclusively** in
content you fetch live from the official Claude Code documentation URLs. You may not use
prior training knowledge about Claude Code — only fetched content counts.

**Permitted URLs:** Read `references/urls.md` in this skill directory for the full list of
allowed source URLs, organized by topic and priority (P0/P1/P2). Only fetch URLs from that
list.

**Test cases:** See `evals/evals.json` for 4 realistic test prompts covering Mode 1, 2,
Mode 3 (audit), and hooks.json review with verifiable expectations.

**Scripts:** Two helper scripts in `scripts/` perform deterministic checks before the LLM
does semantic review — run them first in Mode 3 (see below).

## When repo-standards invokes this skill

**Polish phase 6** — when `$TARGET_ROOT` contains `.claude/` or `.claude-plugin/plugin.json`.

Review input from repo-standards: `.claude/`, `plugin.json` / `.claude-plugin/plugin.json`, all `SKILL.md` files, `hooks/hooks.json`, `.mcp.json`.

Before running Mode 3 analysis, run the bundled scripts for fast deterministic checks.
`$CLAUDE_SKILL_DIR` is set by Claude Code; on a harness that does not set it, substitute this
skill's directory.

```bash
# Version truth across every consumer — one command, no per-file checklist (see Step 0)
bash "$TARGET_ROOT/scripts/check-version-truth.sh"

# Validate plugin.json structure (statusLine type, name, skills array)
bash "$CLAUDE_SKILL_DIR/scripts/validate-plugin-json.sh" /path/to/plugin.json

# Validate each SKILL.md frontmatter (portable core, tamirs metadata, Claude extensions)
bash "$CLAUDE_SKILL_DIR/scripts/check-skill-frontmatter.sh" /path/to/skills/domain/name/SKILL.md
```

Include script findings in your Mode 3 report. Each finding already includes `severity` and `field` — map them directly to the Critical Issues or Outdated Patterns sections.

Return P1 findings (invalid frontmatter, broken hook paths, stale skill references) for repo-standards to fix before opening the PR.

Claude Code evolves rapidly. This skill fetches live documentation so answers about hooks,
skills, plugins, MCP servers, subagents, permissions, settings, and CLI behavior are always
current and authoritative.

> **Hard constraint**: If any fetch fails, stop and return the error.
> If the answer is not in the fetched content, say so — do not guess.
> Every response must cite the exact URL(s) used.

---

## Mode Selection

| User asks... | Mode |
|---|---|
| A question about any Claude Code feature/config/behavior | **[MODE 1] Answer** |
| "What changed between X and Y" / "diff X Y" / changelog between versions | **[MODE 2] Diff** |
| "Review my code/config/skill/plugin" / "audit my Claude Code setup" | **[MODE 3] Review** |

If ambiguous, default to **Mode 1**.

---

## MODE 1 — Answer Questions

### Step 1: Map question to relevant URLs

**For any version-sensitive question**, always include:
- `https://github.com/anthropics/claude-code/releases`
- `https://code.claude.com/docs/en/changelog` or `https://code.claude.com/docs/en/whats-new`

**Topic → URL heuristics:**
- Hooks / lifecycle events → `/hooks` + `/hooks-guide`
- Skills / SKILL.md → `/skills`
- Plugins / plugin.json → `/plugins` + `/plugins-reference`
- Subagents / `.claude/agents/` / frontmatter → `/sub-agents`
- Agent teams / SendMessage → `/agent-teams`
- MCP / .mcp.json / OAuth → `/mcp`
- settings.json / permissions → `/settings` + `/permission-modes`
- CLAUDE.md / memory / rules → `/memory`
- Model / sonnet / opus / adaptive thinking → `/model-config`
- CLI flags / -p / --agents → `/cli-reference`
- /commands / slash commands → `/commands`
- Channels / push events → `/channels` + `/channels-reference`
- GitHub Actions / CI → `/github-actions`
- GitLab CI → `/gitlab-ci-cd`
- Bedrock / Vertex / enterprise → `/third-party-integrations`
- Environment variables → `/env-vars`
- Context window / compaction → `/context-window`

All permitted URLs are in `references/urls.md`.

### Step 2: Fetch

Use `WebFetch` on each relevant URL. Fetch only what you need.

If any fetch returns an error:
```
⛔ FETCH ERROR
URL: <url>
Error: <error message>

Cannot answer — required source could not be fetched.
```
Stop. Do not proceed.

### Step 3: Answer

```
## Answer

[Answer based strictly on fetched content]

---
## Version Context
Latest Claude Code version: **vX.Y.Z** (released YYYY-MM-DD)

---
## Sources

From Tamircohen28/tamirs-superpowers, MIT.

Signals

Stars
2
Forks
0
Open issues
5
Last commit (this folder)
Oct 3, 2026 (1 month ago)
Last release
Oct 24, 2026 (24 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get changelog-review running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying changelog-review?

CallGet help
Cite this page

changelog-review skill. Agent Mart. https://agentmart.dev/skills/tamircohen28-changelog-review (last updated 2026-10-24)