All skills
Skillintermediate
Agent Knowledge Library
Index of reference documents for AI agent development, prompt engineering, and platform integration.
Claude Code Knowledge Pack7/10/2026
Overview
Agent Knowledge Library
Index of reference documents for AI agent development, prompt engineering, and platform integration.
General Knowledge
| Topic | File | Key Content |
|---|---|---|
| Agent Architecture | AI-AGENT-ARCHITECTURE-RESEARCH.md | Reasoning patterns (ReAct, CoT, ToT, GoT, Reflexion), state management, agent loops |
| Prompt Engineering | PROMPT-ENGINEERING-REFERENCE.md | System prompts, few-shot, chain-of-thought, structured output, agentic prompting |
| Tool Use & MCP | FUNCTION-CALLING-TOOL-USE-REFERENCE.md | Function schemas, parallel calls, MCP protocol, tool security |
| Context Optimization | CONTEXT-OPTIMIZATION-REFERENCE.md | Token budgeting, RAG, prompt caching, lost-in-middle problem |
| Multi-Agent Systems | MULTI-AGENT-SYSTEMS-REFERENCE.md | Orchestration patterns, frameworks (LangGraph, AutoGen, CrewAI), communication |
| Instruction Following | LLM-INSTRUCTION-FOLLOWING-RELIABILITY.md | Instruction hierarchy, guardrails, hallucination prevention, reliability |
Platform Integration
| Platform | File | Key Content |
|---|---|---|
| Claude Code | CLAUDE-CODE-REFERENCE.md | Hooks, skills, MCP, Agent SDK, subagents, configuration |
| OpenCode | OPENCODE-REFERENCE.md | Commands, agents, MCP integration, project instructions |
| Codex CLI | CODEX-REFERENCE.md | Skills, sandbox, MCP, session management |
Project-Specific
| Topic | File | Key Content |
|---|---|---|
| Workflow Agents | workflow.md | /next-task phases, /ship workflow, agent tool restrictions |
| Release Process | release.md | Version bumping, GitHub Actions, npm publishing |
Topic Ownership
To avoid duplication, each topic has one canonical source:
- Reasoning patterns (ReAct, Plan-and-Execute, ToT) → AI-AGENT-ARCHITECTURE-RESEARCH.md
- Prompting techniques (CoT, few-shot, structured output) → PROMPT-ENGINEERING-REFERENCE.md
- Tool/function calling (schemas, MCP) → FUNCTION-CALLING-TOOL-USE-REFERENCE.md
- Context management (RAG, caching, tokens) → CONTEXT-OPTIMIZATION-REFERENCE.md
- Multi-agent coordination (orchestration, frameworks) → MULTI-AGENT-SYSTEMS-REFERENCE.md
- Reliability (guardrails, hallucinations) → LLM-INSTRUCTION-FOLLOWING-RELIABILITY.md