Menu
Route agentic RAG work across query routing, tool-use retrieval, and multi-agent orchestration.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Use this skill when building, debugging, or improving Retrieval-Augmented Generation systems, including chunking, vector database selection, hybrid search, reranking, multimodal RAG, code documentation RAG, retrieval latency, and production RAG architecture.
Chunk nested documents into parent-child levels so retrieval can move from broad sections to fine-grained passages.
Use semantic boundaries and embedding similarity to chunk text for higher-relevance retrieval.
Route RAG chunking decisions across semantic, hierarchical, sliding-window, contextual-header, and framework-selection strategies.
Use overlapping windows to preserve context across chunk boundaries while controlling retrieval size.
Reduce retrieval latency with caching, batching, and index-level optimization.
| name | rag-agents |
| title | RAG Agents |
| description | Route agentic RAG work across query routing, tool-use retrieval, and multi-agent orchestration. |
| category | rag-agents |
| tags | ["agents","agentic-rag","orchestration","routing"] |
| allowed-tools | ["Read","Grep","Glob"] |
Use this parent skill when a single retrieve-then-generate pass is not enough and the system must decide what to retrieve, which tools to call, or how to split work across agents. Route to the child skill that matches whether the problem is routing decisions or multi-agent orchestration.
Static RAG pipelines answer every query the same way: one retrieval, one generation. Complex questions need conditional retrieval, multiple sources, tool calls, or decomposition into subtasks. Bolting this onto a linear pipeline produces brittle, hard-to-debug systems.
Add agentic control only when queries genuinely vary in required steps or sources. A fixed pipeline is cheaper and easier to evaluate.
Add a router that chooses retrieval strategy or tool per query before reaching for multiple coordinated agents.
Move to planner and worker agents when subtasks are genuinely independent or need specialized handling, and keep each agent narrowly scoped.
