IckyMCP

RAG MCP server for document search across large collections with local embeddings and SQLite storage.

python

0

GitHub Stars

python

Language

7 months ago

First Indexed

3 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "dl1683-ickymcp": {
      "command": "python",
      "args": [
        "/path/to/ickyMCP/run.py"
      ],
      "env": {
        "ICKY_DB_PATH": "/path/to/icky.db",
        "ICKY_CHUNK_SIZE": "4000",
        "ICKY_CHUNK_OVERLAP": "500",
        "ICKY_EMBEDDING_MODEL": "nomic-ai/nomic-embed-text-v1.5"
      }
    }
  }
}

IckyMCP is a local MCP server designed to index and semantically search large document collections. It tokenizes documents into large chunks, computes embeddings locally, and stores everything in a portable SQLite database so you can quickly retrieve relevant passages from PDFs, Word files, slides, Markdown, or plain text.

How to use

You run the IckyMCP server locally and connect with an MCP client to index documents and run semantic searches. Typical workflows include indexing a folder of contracts, then querying for relevant clauses, and refreshing only changed files to keep the index up to date. You can search across indexed documents, find chunks similar to a given text, refresh the index when files change, list what has been indexed, delete outdated documents, and check the server status for health and usage statistics.

How to install

Prerequisites you need before installation are Python 3.10 or newer and a system with enough memory and disk space to handle large document embeddings.

Step 1: Set up a virtual environment and activate it.

# Clone or copy the project
cd ickyMCP

# Create virtual environment
python -m venv venv
source venv/bin/activate  # or `venv\Scripts\activate` on Windows
```,

Step 2: Install dependencies.

pip install -r requirements.txt
```,

Step 3: Alternatively install as a package.

pip install -e .
```,

Configuration and environment variables

Configure the MCP server to control how documents are chunked and embedded, where the database is stored, and which embedding model to use.

Environment variables you can set:

ICKY_CHUNK_SIZE=4000 ICKY_CHUNK_OVERLAP=500 ICKY_DB_PATH=./icky.db ICKY_EMBEDDING_MODEL=nomic-ai/nomic-embed-text-v1.5 `

## Claude code configuration to run the MCP server

Add the MCP server settings to your Claude configuration so the client can start and manage the server.

{ "mcpServers": { "ickyMCP": { "command": "python", "args": ["/path/to/ickyMCP/run.py"], "env": { "ICKY_CHUNK_SIZE": "4000", "ICKY_CHUNK_OVERLAP": "500", "ICKY_DB_PATH": "/path/to/icky.db" } } } } `


## Available tools

### index

Index documents from a file or directory to build the search index.

### search

Perform a semantic search across indexed chunks and retrieve relevant results.

### similar

Find chunks similar to a given text fragment to explore related content.

### refresh

Re-index only files that have changed since the last indexing.

### list

List all documents and chunks currently indexed.

### delete

Remove documents from the index or clear the entire index.

### status

Query server status and usage statistics.