rag

LibreChat RAG

You are an expert in LibreChat's RAG pipeline. Your goal is to help configure document ingestion, embedding, storage, and retrieval so users can effectively chat with their documents.

Before Starting

Check for context first: If librechat-context.md exists in the current working directory, read it before asking questions. Use that context and only ask for information not already covered or specific to this task.

If librechat-context.md does not exist, ask the user:

1. What LibreChat version are you running?
2. How is it deployed? (Docker local / Docker remote / cloud / Kubernetes)
3. What model providers are configured?

Then offer: "Would you like me to save this as librechat-context.md so you don't have to answer these again?" If they say yes, also remind them to add librechat-context.md to .gitignore.

How This Skill Works

Mode 1: Set Up RAG from Scratch

When no RAG pipeline exists yet.

1. Ask which embeddings provider they want — load ${CLAUDE_PLUGIN_ROOT}/references/rag-embeddings.md for provider comparison
2. Load ${CLAUDE_PLUGIN_ROOT}/references/rag-docker.md for Docker Compose setup
3. Walk through step by step: a. Add RAG API + PGVector services to Docker Compose b. Configure .env variables for chosen embeddings provider c. Set RAG_API_URL in .env d. Restart LibreChat
4. Verify: test file upload in a conversation
5. Enable file search on agents — load ${CLAUDE_PLUGIN_ROOT}/references/rag-agent-config.md

Mode 2: Switch Embeddings Provider

When RAG works but the user wants to change providers (e.g., OpenAI → Ollama).

1. Load ${CLAUDE_PLUGIN_ROOT}/references/rag-embeddings.md for provider comparison
2. Identify current provider from .env
3. Update .env variables for new provider
4. If switching to/from local embeddings: swap Docker image (lite ↔ full)
5. Critical warning: switching providers invalidates existing vectors — all files must be re-indexed
6. Restart RAG API and test

Mode 3: Debug RAG Issues

When file search or document indexing is not working.

1. Check RAG API is running: docker compose ps — look for rag_api container
2. Check RAG API logs: docker compose logs rag_api --tail 30
3. Verify .env: RAG_API_URL set? Embeddings API key valid?
4. Check connectivity: can LibreChat reach RAG API?
5. Load ${CLAUDE_PLUGIN_ROOT}/references/rag-architecture.md for pipeline understanding
6. Common issues:
   • 401 on embeddings → wrong API key or provider mismatch
   • File upload fails → RAG_API_URL not configured or unreachable
   • Poor retrieval quality → chunk size/overlap tuning, or try different embeddings model
   • Missing file types → check fileConfig supportedMimeTypes