Knowledge Base

This MCP server provides tools for listing and retrieving content from different knowledge bases.

typescript

GitHub Stars

typescript

Language

6 months ago

First Indexed

2 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": {
    "jeanibarz-knowledge-base-mcp-server": {
      "command": "node",
      "args": [
        "/path/to/knowledge-base-mcp-server/build/index.js"
      ],
      "env": {
        "LOG_FILE": "/var/log/knowledge-base.log",
        "LOG_LEVEL": "info",
        "OLLAMA_MODEL": "dengcao/Qwen3-Embedding-0.6B:Q8_0",
        "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY",
        "OLLAMA_BASE_URL": "http://localhost:11434",
        "FAISS_INDEX_PATH": "/home/user/knowledge_bases/.faiss",
        "OPENAI_MODEL_NAME": "text-embedding-ada-002",
        "EMBEDDING_PROVIDER": "ollama",
        "HUGGINGFACE_API_KEY": "YOUR_HUGGINGFACE_API_KEY",
        "HUGGINGFACE_MODEL_NAME": "sentence-transformers/all-MiniLM-L6-v2",
        "KNOWLEDGE_BASES_ROOT_DIR": "/path/to/knowledge_bases"
      }
    }
  }
}

You can run a Knowledge Base MCP Server that indexes text content from multiple knowledge bases and provides fast, semantic search results through an MCP client. It supports local embeddings via Ollama or cloud embeddings via OpenAI or HuggingFace, automatically builds a FAISS index, and updates it when knowledge base files change, making it easy to retrieve relevant chunks for your queries.

How to use

To use the Knowledge Base MCP Server, you will run one or more MCP configurations that expose the server to your MCP client. You can run a local stdio-based MCP process that talks to the server binary via node, and you can configure multiple embedding providers (ollama, openai, huggingface). The server exposes two core capabilities: listing available knowledge bases and retrieving semantically similar chunks from the knowledge bases. When you call retrieve functionality, you can limit the search to a specific knowledge base or search across all configured bases, and you will receive up to ten matching chunks by default, each with a similarity score.

How to install

Prerequisites you need before installation:

• Node.js 16 or higher and npm installed on your system.

Installation steps cover both automatic and manual approaches. Choose the method you prefer and follow the steps.

Additional configuration and usage notes

The server supports three embedding providers. Choose one and set the corresponding environment variables.

Troubleshooting

If you encounter issues with indexing or FAISS permissions, verify write access to the FAISS index path and to the .index directories within each knowledge base. Enable structured logging by setting a log file path and adjust log levels as needed to diagnose problems.

Available tools

list_knowledge_bases

Lists all configured knowledge bases available for semantic search.

retrieve_knowledge

Retrieves similar content chunks from the knowledge bases based on a query. Can be restricted to a specific KB or applied to all available bases; returns up to 10 chunks by default with similarity scores.

Built by

VeilStrat

AI signals for GTM teams