KB MCP Server

v1.0.0Knowledge & Memorystable

An MCP server aimed to be portable, local, easy and convenient to support semantic/graph based retrieval of txtai 'all in one' embeddings database. Any txtai embeddings db in tar.gz form can be loaded

kbmcpai-integration
Share:
68
Stars
0
Downloads
0
Weekly
0/5

What is KB MCP Server?

KB MCP Server is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to mcp server aimed to be portable, local, easy and convenient to support semantic/graph based retrieval of txtai 'all in one' embeddings database. any txtai embeddings db in tar.gz form can be loaded

An MCP server aimed to be portable, local, easy and convenient to support semantic/graph based retrieval of txtai 'all in one' embeddings database. Any txtai embeddings db in tar.gz form can be loaded

This server falls under the Knowledge & Memory category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • An MCP server aimed to be portable, local, easy and convenie

Use Cases

Load portable txtai embeddings databases locally.
Perform semantic and graph-based knowledge retrieval.
Search through custom knowledge bases efficiently.
Geeksfino

Maintainer

LicenseMIT License
Languagepython
Versionv1.0.0
UpdatedMay 15, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx kb

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use KB MCP Server

The KB MCP server provides AI assistants with semantic and graph-based retrieval over locally stored knowledge bases built on the txtai 'all-in-one' embeddings format. You can load any txtai embeddings database — packaged as a directory or compressed .tar.gz archive — and instantly query it through your MCP client using natural language. It supports hybrid search combining vector similarity and keyword matching, causal boosting for 'why/how' questions, and knowledge graph traversal, making it well-suited for private document collections, internal wikis, and research corpora.

Prerequisites

  • Python 3.10 or later (3.10 recommended for compatibility)
  • uv package manager (pip install -U uv) or conda environment manager
  • A pre-built txtai embeddings database (directory or .tar.gz file) — build one with the kb-build command
  • An MCP client such as Claude Desktop, Cursor, or any stdio-compatible client
1

Create a Python virtual environment

Use uv to create an isolated environment with Python 3.10. Activate it before proceeding.

pip install -U uv
uv venv --python=3.10
source .venv/bin/activate
2

Install kb-mcp-server

Install the kb-mcp-server package from PyPI into your virtual environment.

uv pip install kb-mcp-server
3

Build a knowledge base from your documents

Use the kb-build utility to index your documents and export them as a portable .tar.gz archive. Provide a config.yml to control chunking, embedding model, and other options.

kb-build --input /path/to/documents --export knowledge_base.tar.gz
4

Verify the knowledge base with a test search

Before connecting to an MCP client, confirm the knowledge base loads correctly and returns relevant results.

kb-search knowledge_base.tar.gz "What is the return policy?" --limit 5
5

Configure your MCP client

Add the kb-mcp-server to your claude_desktop_config.json or equivalent config file, pointing to your knowledge base file.

6

Enable causal boosting (optional)

For better answers to 'why' and 'how' questions, start the server with the --enable-causal-boost flag or provide a custom boost configuration YAML.

kb-mcp-server --embeddings knowledge_base.tar.gz --enable-causal-boost

KB MCP Server Examples

Client configuration

Add this to claude_desktop_config.json. The TXTAI_EMBEDDINGS environment variable can also be used instead of the --embeddings argument.

{
  "mcpServers": {
    "kb": {
      "command": "uvx",
      "args": [
        "--from", "kb-mcp-server",
        "kb-mcp-server",
        "--embeddings", "/path/to/knowledge_base.tar.gz"
      ]
    }
  }
}

Prompts to try

Once the server is loaded with your knowledge base, query it using natural language through your AI assistant.

- "Search the knowledge base for information about onboarding procedures"
- "Find documents related to the Q3 product roadmap"
- "Why did the project fail according to the post-mortem?"
- "Show me everything connected to the 'authentication' topic in the knowledge graph"
- "Summarize all documents about customer refund policies"

Troubleshooting KB MCP Server

Server fails to start with 'embeddings not found' error

Make sure the path passed to --embeddings (or TXTAI_EMBEDDINGS) points to an actual directory or .tar.gz file created by kb-build. Relative paths can cause issues — use absolute paths.

Search returns empty or irrelevant results

The embedding model used to build the index must match the model used to query. If you used a custom model in kb-build, specify the same model in your config.yml when running kb-mcp-server.

Python version compatibility errors on install

This package requires Python 3.10 exactly for best compatibility. Use 'uv venv --python=3.10' to pin the version and ensure you are inside the activated venv before installing.

Frequently Asked Questions about KB MCP Server

What is KB MCP Server?

KB MCP Server is a Model Context Protocol (MCP) server that mcp server aimed to be portable, local, easy and convenient to support semantic/graph based retrieval of txtai 'all in one' embeddings database. any txtai embeddings db in tar.gz form can be loaded It connects AI assistants to external tools and data sources through a standardized interface.

How do I install KB MCP Server?

Follow the installation instructions on the KB MCP Server GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with KB MCP Server?

KB MCP Server works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is KB MCP Server free to use?

Yes, KB MCP Server is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Knowledge & Memory MCP Servers

Explore all knowledge & memory servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "kb": { "command": "npx", "args": ["-y", "kb"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use KB MCP Server?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides