Files-DB-MCP Server

Files-DB-MCP provides a local vector search server that indexes code projects and exposes an MCP interface for chat-based coding assistants. It automatically discovers your project structure, watches for changes in real time, and serves fast semantic search over code embeddings, enabling you to ask questions about your codebase and get relevant results from any MCP client.

How to use

You will run the MCP server in any project directory and connect your MCP client to it. Start the service once per project; it will index your files in the background and respond to search queries as you work. Use an MCP-compatible client (such as Claude Code or any tool that speaks MCP) to send search queries, file lookups, or semantic questions about your code.

How to install

Prerequisites you need before installation:

• Docker and Docker Compose are required to run the service.

# Option 1: Clone and Setup (Recommended)
# Using SSH
git clone git@github.com:randomm/files-db-mcp.git ~/.files-db-mcp && bash ~/.files-db-mcp/install/setup.sh

# Using HTTPS
git clone https://github.com/randomm/files-db-mcp.git ~/.files-db-mcp && bash ~/.files-db-mcp/install/setup.sh

# Option 2: Automated Installation Script
curl -fsSL https://raw.githubusercontent.com/randomm/files-db-mcp/main/install/install.sh | bash

After installation, you start the service from within any project directory by running the MCP server command. It will detect your project files, begin indexing in the background, and start responding to MCP queries immediately.

Configuration and runtime notes

The server can operate without configuration, but you can customize its behavior with environment variables to tailor embeddings, startup speed, and what to ignore.

# Example environment variable overrides
EMBEDDING_MODEL=sentence-transformers/all-MiniLM-L6-v2
FAST_STARTUP=true
QUANTIZATION=true
BINARY_EMBEDDINGS=false
IGNORE_PATTERNS=build,dist,.venv

Available tools

zero_config

Auto-detects project structure and applies sensible defaults without manual configuration.

real_time_monitoring

Continuously watches for file changes and updates the index in real time.

vector_search

Performs semantic code search to retrieve relevant snippets and files based on intent.

mcp_interface

Exposes an MCP endpoint compatible with MCP clients such as Claude Code for querying and administration.

model_embedding

Uses open-source code embeddings to represent code for fast semantic comparison.