LLM 與 Agent

自我進化 Agent：給定目標，自主學習並迭代改進直到完成。 觸發詞：evolve、進化、自我學習、迭代改進、達成目標。 適用於：複雜任務自動完成、需要學習新技能的任務、多步驟迭代改進。

Structured code review output format for the reviewer agent. Provides APPROVED/ISSUES format with file:line references. Use when: reviewing code changes, code quality checks, architecture compliance checks, approving code for merge, identifying critical/major/minor issues.

Post-PRD project initialization skill for the /initialize-project command.Guides constitution customization, agent training, and workflow configurationbased on completed Product Requirements Document. Ensures framework istailored to specific project needs while maintaining constitutional compliance.

Investigate platform incidents, perform RCA, create incident documentation, and follow alert runbooks in the Kagenti platform

Analyze and improve Claude Code skills against best practices. Supports scope keywords (project/global/all) to target current project skills, ~/.claude/skills/, or both. Default is project-only. Evaluate SKILL.md quality using a 100-point rubric. Activated when users mention "improve skill", "analyze skill", "audit skills", or need to refactor existing skills.

Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.

Assess Claude Code knowledge and route to appropriate learning level.Use when: user wants to learn Claude Code, asks for guidance, or says "teach me".Triggers: "learn Claude Code", "teach me", "I'm new", "where do I start", "beginner".

This skill should be used when validating user prompts, plans, or requests before execution. It conducts independent validation, questions assumptions, references official sources and documentation, identifies potential issues, and presents revised recommendations to ensure accuracy and completeness.

AI self-improvement analyst. Tracks AI agent mistakes, analyzes failure patterns, and proposes system improvements. Implements continuous learning loop for trading system enhancement.

Creates Claude Code hooks for automation and workflow customization.Guides through hook events, configuration, and script creation.Use when user wants to create a hook, automate Claude Code, or asks about hook events.

Self-Improving Agent that monitors all other agent skills, analyzes their logs,detects issues, and proposes improvements.AUTO-TRIGGERS:- Every 30 minutes (scheduled)- When error rate > 5% (any agent)- When 3+ recurring errors in 24h (same error type)- When performance degrades > 2x baseline

Build AI chat interfaces with custom backends, authentication, and context injection.Use when integrating chat UI with AI agents, adding auth to chat, injecting user/page context,or implementing httpOnly cookie proxies. Covers ChatKitServer, useChatKit, and MCP auth patterns.NOT when building simple chatbots without persistence or custom agent integration.

Use when working with Qdrant vector database for semantic search and RAG. Covers collection setup, embedding generation, vector upsert/search, HNSW indexing, filtering, and integration with OpenAI embeddings for textbook content retrieval.

Use when spec is complete and you need detailed implementation tasks for LLM agents. Creates execution plans with exact file paths, complete code examples, and verification steps. Triggers: 'write plan', 'execution plan', 'implementation plan', 'break down into tasks', 'detailed steps'.

This skill should be used when agents need to provide clickable code references to users in reports, investigations, or bug findings. It ensures file references use the simple path:line_number format that Claude Code automatically converts to clickable links. Do not use for general documentation or conceptual explanations.

Automatiza processo de pesquisa de documentação oficial, extração de melhores práticas e aplicação no projeto Plataforma B2B de treinamento técnico corporativo através de skills/agents/hooks

Guided workflow for creating new Claude Code Skills with proper structure, documentation, templates, subagents, and hooks. Use this when the user wants to create a new skill or asks you to build tooling/automation. Ensures skills follow official conventions and are production-ready.

Store and retrieve documents using ChromaDB for RAG workflows. Use for persistent memory, research storage, and semantic search.

Comprehensive guide for implementing Mapbox GL JS maps in Angular with ngx-mapbox-gl. Use when building interactive maps, adding markers/popups/controls, implementing geolocation, creating coverage zones with polygons, working with GeoJSON layers, geocoding/directions, or managing map state. Covers installation, components, common patterns, and advanced features for Angular standalone components with signals.

Creates a battle-tested ChatGPT-style chatbot widget that solves real-world production issues. Features infinite re-render protection, text selection "Ask AI", RAG backend integration, streaming SSE, and comprehensive performance monitoring.