abstract

Generate LEARNINGS.md from skill execution logs.

Analyze individual skill complexity and get modularization recommendations. For ecosystem-wide evaluation, use /skills-eval instead.

Harden skills against rationalization and bypass behaviors

Generate context optimization report for skill directories

Create slash commands with brainstorming and best practices

Create hooks with brainstorming and security-first design

Create skills with brainstorming and TDD methodology

Manually evaluate a recent skill execution for qualitative feedback.

Evaluate all hooks in a plugin for quality and compliance

Automatically identify and implement skill improvements based on execution logs and user evaluations.

Analyze and enhance Makefiles for complete functionality coverage with auto-generation capability

Rigorous review of plugin architecture quality. Orchestrates multiple evaluation tools to provide a unified health report covering skills, commands, hooks, agents, token efficiency, and bloat detection.

Check GitHub Discussions for highly-voted learnings and promote them to Issues.

Evaluate Claude Code rules in .claude/rules/ directories for quality

Evaluate skills for quality, compliance, and improvements

Test skills through RED/GREEN/REFACTOR TDD phases

Validate hooks for security, performance, and SDK compliance

Validate plugin structure, schema, and naming. Use for plugin creation, debugging, or verification.

{

Validates Claude Code plugin structure against official requirements - checks plugin.json schema, verifies referenced paths exist, validates kebab-case naming, and validates skill frontmatter is complete.

{

Facilitates human-in-loop qualitative evaluation of skill executions.

Implements skill improvements based on observability data from LEARNINGS.md. Prioritizes by frequency × impact / ease, generates proposals, validates changes.

NEVER escalate without investigation first. This is the Iron Law. Use when evaluating whether to escalate models, facing genuine complexity requiring deeper reasoning, novel patterns with no existing solutions, high-stakes decisions requiring capability investment. Do not use when thrashing without investigation - investigate root cause first. DO NOT use when: time pressure alone - urgency doesn't change task complexity. DO NOT use when: "just to be safe" - assess actual complexity instead.

Use this skill BEFORE writing any hook. Check even if unsure. Use when creating new hooks for tool validation, logging operations for audit, injecting context before prompts, enforcing project-specific workflows, preventing dangerous operations in production. Do not use when logic belongs in core skill - use Skills instead. DO NOT use when: complex multi-step workflows needed - use Agents instead. DO NOT use when: behavior better suited for custom tool.

Use this skill BEFORE deploying hooks to production. Use when auditing existing hooks for security vulnerabilities, benchmarking hook performance, implementing hooks using Python SDK, understanding hook callback signatures, validating hooks against compliance standards. Do not use when deciding hook placement - use hook-scope-guide instead. DO NOT use when: writing hook rules from scratch - use hook-authoring instead. DO NOT use when: validating plugin structure - use validate-plugin instead.

Surfaces expert frameworks and proven methodologies before creating OR evaluating skills, hooks, agents, or commands. Helps select approaches from domain masters. Use when starting creation workflows, evaluating methodology gaps, or seeking domain expertise. Do not use if you already have a specific methodology or are fixing syntax/structural issues.

Architect skills as modular blocks to control token usage and complexity. Use when creating skills >150 lines, breaking down monolithic skills, or planning new architecture. Do not use for evaluating existing skills (use skills-eval) or writing human-facing prose (use writing-clearly-and-concisely). Check this skill before starting any new skill development.

Evaluate and validate Claude Code rules in .claude/rules/ directories. Use when auditing rule file quality, validating frontmatter and glob patterns, or checking rules organization before deployment. Do not use when writing new rules from scratch - use rule authoring guides instead. Do not use when evaluating skills or hooks - use skills-eval or hooks-eval instead.

Reference these patterns to validate consistency across the ecosystem. Use when creating new skills or hooks that need consistent patterns, implementing validation logic, setting up error handling, creating test scaffolding, referencing standard workflow structures. Do not use when pattern is specific to one skill only. DO NOT use when: pattern is still evolving - wait for stability. DO NOT use when: pattern is context-dependent requiring variations.

Guide to effective Claude Code skill authoring using TDD methodology and persuasion principles. Use when creating new skills, improving compliance, or validating quality before deployment. Do not use for evaluating existing skills (use skills-eval) or analyzing architecture (use modular-skills). Follow the Iron Law: write a failing test before writing any skill.

Evaluate and improve Claude skill quality through auditing. Use when reviewing skill quality, preparing skills for production, or auditing existing skills. Do not use when creating new skills (use modular-skills) or writing prose (use writing-clearly-and-concisely). Use this skill before shipping any skill to production.

TDD-style testing methodology for skills using fresh subagent instances to prevent priming bias and validate skill effectiveness. Use when validating skill improvements, testing skill effectiveness, preventing priming bias, measuring skill impact on behavior. Do not use when implementing skills (use skill-authoring instead), creating hooks (use hook-authoring instead).

4 hooks across 3 events