Quant

What it does

Quant is a lightweight RAG (Retrieval-Augmented Generation) index that turns a local folder into a searchable knowledge base. It watches the filesystem in real-time, chunks content, and uses Ollama for embeddings, allowing AI agents to perform high-quality semantic searches over your local documents and code.

Tools

• search: Performs hybrid semantic and keyword search over indexed content.
• list_sources: Lists all documents currently tracked in the index.
• index_status: Returns statistics on document count, DB size, and embedding state.
• find_similar: Locates chunks most similar to a specific chunk ID.
• drill_down: Explores a topic by finding diverse related chunks across different files.
• summarize_matches: Provides a high-level overview of all documents matching a query.
• list_collections: Lists named document collections.
• delete_collection: Removes all data associated with a specific collection.

Installation

For macOS/Linux:

curl -fsSL https://raw.githubusercontent.com/koltyakov/quant/main/scripts/install.sh | sh

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "quant": {
      "command": "quant",
      "args": ["mcp", "--dir", "/path/to/your/docs"]
    }
  }
}

Supported hosts

• claude
• codex
• cursor