aio-rag-kit

RAG Kit Skill

Vector database operations (Qdrant) for Retrieval-Augmented Generation via nguyenvanduocit/rag-kit.

Environment

• Go: !which go 2>/dev/null || echo "NOT INSTALLED"
• rag-kit: !which rag-kit 2>/dev/null || echo "NOT INSTALLED"
• rag-cli: !which rag-cli 2>/dev/null || echo "NOT INSTALLED"
• QDRANT_HOST: !echo ${QDRANT_HOST:-NOT SET}
• QDRANT_PORT: !echo ${QDRANT_PORT:-NOT SET}
• QDRANT_API_KEY: ![ -n "$QDRANT_API_KEY" ] && echo "SET" || echo "NOT SET"
• OPENAI_API_KEY: ![ -n "$OPENAI_API_KEY" ] && echo "SET" || echo "NOT SET"
• MCP configured: !cat .mcp.json 2>/dev/null | grep -q rag && echo "YES" || echo "NO"

Install (skip if already installed above)

Prerequisites: Qdrant running (local or cloud) + OpenAI API key for embeddings.

# Run Qdrant locally
docker run -d --name qdrant -p 6333:6333 -p 6334:6334 qdrant/qdrant

# Install binaries
go install github.com/nguyenvanduocit/rag-kit@latest
go install github.com/nguyenvanduocit/rag-kit/cmd/rag-cli@latest

Or use Qdrant Cloud: https://cloud.qdrant.io/

Add to .mcp.json:

{
  "mcpServers": {
    "rag": {
      "command": "rag-kit",
      "env": {
        "QDRANT_HOST": "localhost",
        "QDRANT_PORT": "6333",
        "QDRANT_API_KEY": "",
        "OPENAI_API_KEY": "sk-xxxxxxxxxxxxxxxxxxxx"
      }
    }
  }
}

Optional: ENABLE_TOOLS — comma-separated list to restrict available tool groups. Restart Claude Code after configuring.

MCP Tools (prefix: rag_)

Collection Management

Tool	Usage
rag_create_collection	(collection_name: "docs", vector_size: 1536) — use 1536 for OpenAI text-embedding-3-small
rag_list_collections	()
rag_delete_collection	(collection_name: "docs")

Content Indexing

rag_index_content(
  collection_name: "docs",
  content: "Document text to index...",
  metadata: {"source": "readme.md", "section": "introduction"}
)
rag_delete_index(collection_name: "docs", point_id: "abc123")