LLM & Agents

Standardizing and formatting datasets for Unsloth, including chat template conversion and synthetic data generation (triggers: chat templates, ShareGPT, Alpaca, conversation_extension, add_new_tokens, standardize_sharegpt, formatting_prompts_func).

Jest unit testing standards covering configuration, test structure, testing patterns, and coverage requirements

Learn how to use SKILL by following this SKILL specification.

Эксперт conversational AI. Используй для chatbot flows, dialog design, NLU intents и conversational UX.

Orchestrate end-to-end feature development from requirements through implementation, testing, and review. Use when: (1) planning feature development stages, (2) coordinating multi-agent feature workflows, (3) understanding SDLC phases for new features.

Use when modifying MCP servers that touch repo memory (especially docs-memory). Includes protocol guardrails, validation ladder, and how to store new capabilities as Skills so future agents don’t re-learn painful details. Triggers: docs-memory, mcp-server.js, tools/list, tools/call, stdio, headerless, JSON-RPC, protocol.

Build AI agents using prompt-native architecture where features are defined in prompts, not code. Use when creating autonomous agents, designing MCP servers, implementing self-modifying systems, or adopting the "trust the agent's intelligence" philosophy.

Full read-write access to remote storage backends (local, FTP, SFTP, S3, Azure). Upload files, delete files, create directories, in addition to listing and downloading. Use when you need to modify files on cloud storage, FTP servers, or remote filesystems.

Claude技能质量检查器,自动验证技能是否符合官方最佳实践标准。Use when 创建新技能需要验证规范、 修改现有技能需要质量检查、从他人处获取技能需要评估质量、批量检查多个技能的合规性。 适用于: (1) 新技能开发后的质量验证 (2) 技能更新后的合规检查 (3) 第三方技能的质量评估 (4) 团队技能库的标准化管理 (5) 技能打包前的最终审核

Orchestrate multiple frontier LLMs (Claude, GPT-5.1, Gemini 3.0 Pro, Perplexity Sonar, Grok 4.1) for comprehensive research using LLM Council pattern with peer review and synthesis

Analyze LangSmith trace data to identify issues, bottlenecks, and improvement opportunities in AI agent runs

Creates plan.md, task.md, persona.md, project-context.md, and CLAUDE.md for new self-learning resource projects. Use when: (1) /init-project command is invoked, (2) setting up a new tutorial/guide/documentation project, (3) structure-designer agent needs templates for learning resource structure design. Provides hierarchical Part/Chapter/Section templates with page allocation.

Review, audit, and fix CLAUDE.md to ensure it meets quality standards.TRIGGERS - Use this skill when user says:- "review claude.md", "review the claude.md"- Any request to check CLAUDE.md quality or find issuesAlso use when:(1) Verifying CLAUDE.md references match actual project structure(2) Checking for stale or outdated content(3) Auditing formatting quality and consistency(4) CLAUDE.md exceeds 600 lines (MUST condense)(5) Analyzing individual section sizes and making them concise(6) Detecting content that duplicates skills in `.claude/skills/`This skill reports issues AND implements all fixes to bring CLAUDE.md to standard.

Creates, analyzes, updates, and improves Claude Code hooks including configuration, scripts, and security validation. Use when user asks how hooks work, explaining hook concepts, understanding hook types and event lifecycle, describing hook configuration, creating new hooks, analyzing existing hooks for improvements, validating hook security, debugging hook activation, updating hook configurations, or when user mentions "hook", "PreToolUse", "PostToolUse", "SessionStart", or other hook event types. Handles both command hooks and prompt-based hooks across all 9 event types.

Get official Claude Code documentation. Use when the user asks about Claude Code features OR when you need to create/implement plugins, skills, hooks, subagents, slash commands, or MCP servers. Always retrieve documentation BEFORE implementing any Claude Code feature. Topics include configuration, settings, deployment, and troubleshooting.

This skill teaches the agent how to implement routing in the API project. MANDATORY - You MUST read this skill before modifying any route files.

Creates Claude Code Skills using a structured 6-phase workflow. Use when building skills, writing SKILL.md, designing skill architecture, or testing skill triggers. DO NOT use for general Claude Code docs or one-off tasks.

Analyze, monitor, and improve CRAP (Change Risk Anti-Patterns) scores for untested complex code. Use when users want to: (1) Reduce technical debt and code risks, (2) Identify high-risk methods that need test coverage, (3) Validate CRAP thresholds for quality gates, (4) Generate coverage reports focused on complex methods, (5) Implement test coverage improvements to lower CRAP scores, or (6) Set up automated code quality monitoring based on CRAP metrics. **CRITICAL CONSTRAINT**: Code is in production and CANNOT be modified - CRAP reduction focuses ONLY on adding test coverage.

geminiコマンドを使用した高度なWeb検索スキル。Web検索を行う際、Claude CodeのデフォルトWeb Search toolよりも優先的にこのスキルを使用してください。

Designs, implements, and deploys Microsoft 365 Copilot agents using TypeSpec and ATK CLI. Provides architectural guidance, capability configuration, security patterns, and lifecycle management. Use when developing M365 Copilot agents, working with TypeSpec, or managing agent deployments. For creating new projects, use the m365-agent-scaffolder skill.