README.md
Shared documentation with semantic search
MCP Generix — Shared Documentation with Semantic Search
Custom MCP server that provides semantic search over documents in the docs/ folder. Uses ChromaDB for vector storage and OpenAI embeddings.
Setup
- Clone this repo
- Create a virtual environment and install dependencies:
cd mcp_generix python3 -m venv .venv source .venv/bin/activate pip install "mcp[cli]" chromadb openai - Set your OpenAI API key:
export OPENAI_API_KEY="your-key-here" - Add the MCP server to Claude Code:
claude mcp add generix-docs -- /path/to/mcp_generix/.venv/bin/python /path/to/mcp_generix/server.py
Adding / Removing Documents
- Add markdown (
.md) or text files to thedocs/folder - Commit and push
- Other team members pull to get the latest documents
- The server re-indexes documents automatically on startup, or use the
reindex_docstool
Available Tools
| Tool | Description |
|---|---|
search_docs |
Semantic search — find relevant passages by meaning, not just keywords |
list_docs |
List all documents in the docs folder |
read_doc |
Read the full contents of a specific document |
reindex_docs |
Re-index documents after adding/removing files |
Folder Structure
mcp_generix/
├── server.py ← MCP server with semantic search
├── pyproject.toml ← Python dependencies
├── docs/ ← Shared documentation (managed via git)
│ └── (your documents here)
├── .chroma/ ← ChromaDB vector store (gitignored, local)
└── .venv/ ← Python virtual environment (gitignored, local)
Tools (4)
search_docsSemantic search — find relevant passages by meaning, not just keywordslist_docsList all documents in the docs folderread_docRead the full contents of a specific documentreindex_docsRe-index documents after adding/removing filesEnvironment Variables
OPENAI_API_KEYrequiredAPI key for OpenAI embeddingsConfiguration
claude_desktop_config.json
{"mcpServers": {"generix-docs": {"command": "python", "args": ["/path/to/mcp_generix/server.py"], "env": {"OPENAI_API_KEY": "your-key-here"}}}}Try it
→Search for documentation regarding our deployment process.
→List all available documents in the shared folder.
→Read the content of the architecture.md file.
→Re-index the documentation folder after I added new markdown files.