LLM & Agents

Implement hooks for permission control and security in custom agents. Use when adding security controls, blocking dangerous operations, implementing audit trails, or designing permission governance.

A meta-skill for documenting and creating reusable skills for Claude Code agents. Use when you discover a technique, pattern, or workflow worth documenting for reuse across projects.

Miyabi tmux マルチエージェント・オーケストレーションとiTerm2プロファイル統合。Agent別カラーテーマ、状態監視、自動プロファイル切り替えを提供。

Propose code changes to the supervisor Claude. Generate diffs and rationale, but do not execute. Wait for supervisor approval.

Synthesize consensus implementation plan from multi-agent debate reports using external AI review

Creates multiple git worktrees and launches parallel Claude sessions for simultaneous task execution. Maximum 3 tasks.

Central authority for Claude Code slash commands. Covers built-in commands, custom slash commands (project and personal), plugin commands, MCP slash commands, SlashCommand tool, frontmatter configuration, arguments ($ARGUMENTS, $1, $2), bash execution, file references, namespacing, and skills vs slash commands comparison. Assists with creating custom commands, configuring command behavior, understanding command types, and troubleshooting command issues. Delegates 100% to docs-management skill for official documentation.

Assist in generating comprehensive plans from templates using extended thinking. Use when generating plans from meta-prompt templates, reviewing plan quality, or ensuring plans meet team standards.

Expert skill for integrating cloud AI APIs (Claude, GPT-4, Gemini). Covers secure API key management, prompt injection prevention, rate limiting, cost optimization, and protection against data exfiltration attacks.

Domain-specific language for categorical prompt composition with functor combinators and natural transformation operators. Use when building composable prompt templates, implementing typed prompt algebras, creating reusable prompt patterns with categorical structure, or designing prompt DSLs that preserve composition properties.

Use when creating releases for Claude Code plugin marketplaces. Supports independent plugin versioning - each plugin can be released separately. Triggered by "release", "bump version", "create release", "publish plugin".

Use when creating llmring.lock file for new project (REQUIRED for all applications), configuring model aliases with semantic task-based names, managing environment-specific profiles (dev/staging/prod), or setting up fallback models - lockfile creation is mandatory first step, bundled lockfile is only for llmring tools

TypeScript type safety patterns and best practices for maximum type coverage. Triggers: 型安全, type safety, any, unknown, 型推論, 型ガード, type guard, discriminated union, 判別可能なUnion, strictNullChecks, 型定義, 型カバレッジ, TypeScript, 暗黙のany, implicit any, 型アサーション, type assertion.

Run Claude Code programmatically without interactive UI. Triggers on: headless, CLI automation, --print, output-format, stream-json, CI/CD, scripting.

Design user-facing agent experts for adaptive UX and personalization. Use when building product features that learn from user behavior, creating per-user expertise files, or implementing AI-driven personalization.

Comprehensive template and guidelines for documenting Field Agents including technical specifications, system prompts, tool specifications, user interactions, and standardized documentation structure

Single source of truth and librarian for ALL Duende Software documentation. Manages local documentation storage, scraping, discovery, and resolution. Use when finding, locating, searching, or resolving Duende documentation; discovering docs by keywords, category, tags, or natural language queries; scraping from llms-full.txt; managing index metadata (keywords, tags, aliases); or rebuilding index from filesystem. Run scripts to scrape, find, and resolve documentation. Handles doc_id resolution, keyword search, natural language queries, category/tag filtering, alias resolution, llms-full.txt parsing, markdown subsection extraction for internal use, hash-based drift detection, and comprehensive index maintenance.

Set up PITER framework elements for AFK agent systems. Use when configuring prompt input sources, triggers, environments, and review processes for autonomous agent workflows.

Three-phase task execution framework with refinement strategies androle-specific prompting. Automatically activates when designing prompts,optimizing task flows, or implementing iterative improvement patterns.

Advises on choosing the right Claude Code extensibility approach (Skills, Commands, Subagents, Hooks, Plugins, MCP, Output Styles). Activates when user discusses extending Claude Code, creating automation, or is unsure which feature to use. Analyzes use cases and recommends appropriate solutions with trade-off explanations. Use when user mentions "should I use", "what's the best way", "skill or command", "which approach", or discusses extensibility options.