knowledge-mcp: Specialized Knowledge Bases for AI Agents

1. Overview and Concept

knowledge-mcp is a MCP server designed to bridge the gap between specialized knowledge domains and AI assistants. It allows users to create, manage, and query dedicated knowledge bases, making this information accessible to AI agents through an MCP (Model Context Protocol) server interface.

The core idea is to empower AI assistants that are MCP clients (like Claude Desktop or IDEs like Windsurf) to proactively consult these specialized knowledge bases during their reasoning process (Chain of Thought), rather than relying solely on general semantic search against user prompts or broad web searches. This enables more accurate, context-aware responses when dealing with specific domains.

Key components:

• CLI Tool: Provides a user-friendly command-line interface for managing knowledge bases (creating, deleting, adding/removing documents, configuring, searching).
• Knowledge Base Engine: Leverages LightRAG to handle document processing, embedding, knowledge graph creation, and complex querying.
• MCP Server: Exposes the search functionality of the knowledge bases via the FastMCP protocol, allowing compatible AI agents to query them directly.

2. About LightRAG

This project utilizes LightRAG (HKUDS/LightRAG) as its core engine for knowledge base creation and querying. LightRAG is a powerful framework designed to enhance Large Language Models (LLMs) by integrating Retrieval-Augmented Generation (RAG) with knowledge graph techniques.

Key features of LightRAG relevant to this project:

• Document Processing Pipeline: Ingests documents (PDF, Text, Markdown, DOCX), chunks them, extracts entities and relationships using an LLM, and builds both a knowledge graph and vector embeddings.
• Multiple Query Modes: Supports various retrieval strategies (e.g., vector similarity, entity-centric, relationship-focused, hybrid) to find the most relevant context for a given query.
• Flexible Storage: Can use different backends for storing key-value data, vectors, graph information, and document status (this project uses the default file-based storage).
• LLM/Embedding Integration: Supports various providers like OpenAI (used in this project), Ollama, Hugging Face, etc.

By using LightRAG, knowledge-mcp benefits from advanced RAG capabilities that go beyond simple vector search.

3. Installation

Ensure you have Python 3.12 and uv installed.

1. Running the Tool: After installing the package (e.g., using uv pip install -e .), you can run the CLI using uvx:

    # General command structure (use either --config or --base)
    uvx knowledge-mcp --config <path-to-your-config.yaml> <command> [arguments...]
    uvx knowledge-mcp --base <path-to-dir-containing-config.yaml> <command> [arguments...]

    # Example: Start interactive shell
    uvx knowledge-mcp --config <path-to-your-config.yaml> shell

1. Configure MCP Client: To allow an MCP client (like Claude Desktop or Windsurf) to connect to this server, configure the client with the following settings. Replace the config path with the absolute path to your main config.yaml.

    {
      "mcpServers": {
        "knowledge-mcp": {
          "command": "uvx",
          "args": [
            "knowledge-mcp",
            "--config",
            "<absolute-path-to-your-config.yaml>",
            "mcp"
          ]
        }
      }
    }

MCP Client Configuration (Docker) For clients like Claude Desktop, Cursor, Windsurf, etc.:

Spawn new container per request (simplest):

    {
      "mcpServers": {
        "knowledge-mcp": {
          "command": "docker",
          "args": [
            "run",
            "--rm",
            "-v",
            "/Users/yourusername/kb:/app/kb",
            "knowledge-mcp"
          ]
        }
      }
    }

This runs knowledge-mcp --base /app/kb mcp inside the container.

Persistent server (better for frequent use):

    docker run -d --name knowledge-mcp-server -v ~/kb:/app/kb knowledge-mcp

Then configure your client to connect to the running instance if it supports that.

1. Set up configuration:

• Copy config.example.yaml to config.yaml.
• Copy .env.example to .env.
• Edit config.yaml and .env to add your API keys (e.g., OPENAI_API_KEY) and adjust paths or settings as needed. The knowledge_base.base_dir in config.yaml specifies where your knowledge base directories will be created.

Docker Setup & Usage

#### Build the image

docker build -t knowledge-mcp .

#### Quick commands (with mounted knowledge base) Start persistent MCP server:

docker run -d --name knowledge-mcp -v ~/kb:/app/kb knowledge-mcp

Run one-off commands:

docker run -it --rm -v ~/kb:/app/kb knowledge-mcp shell
docker run -it --rm -v ~/kb:/app/kb knowledge-mcp create my-notes
docker run --rm -v ~/kb:/app/kb knowledge-mcp list

#### Using docker-compose (recommended) Edit docker-compose.yml to point the volume to your real folder, then:

docker compose up -d
docker compose down
docker compose logs -f

docker compose run --rm knowledge-mcp create work-kb
docker compose run --rm knowledge-mcp shell

#### Required host setup

mkdir -p ~/kb
cp config.example.yaml ~/kb/config.yaml
cp .env.example ~/kb/.env

Then edit:

• ~/kb/config.yaml → set knowledge_base.base_dir: /app/kb
• ~/kb/.env → add your OPENAI_API_KEY

4. Configuration

Configuration is managed via YAML files.

When using Docker, set knowledge_base.base_dir: /app/kb in your config.yaml (the one you mount from your host).

1. Main Configuration (`config.yaml`): Defines global settings like the knowledge base directory (knowledge_base.base_dir), LightRAG parameters (LLM provider/model, embedding provider/model, API keys via ${ENV_VAR} substitution), and logging settings. Refer to config.example.yaml for the full structure and available options.

    knowledge_base:
      base_dir: ./kbs

    lightrag:
      llm:
        provider: "openai"
        model_name: "gpt-4.1-nano"
        api_key: "${OPENAI_API_KEY}"
        # ... other LLM settings
      embedding:
        provider: "openai"
        model_name: "text-embedding-3-small"
        api_key: "${OPENAI_API_KEY}"
        # ... other embedding settings
      embedding_cache:
        enabled: true
        similarity_threshold: 0.90

    logging:
      level: "INFO"
      # ... logging settings

    env_file: .env # path to .env file

1. Knowledge Base Specific Configuration (`<base_dir>/<kb_name>/config.yaml`): Contains parameters specific to querying that knowledge base, such as the LightRAG query mode (default: "hybrid"), top_k results (default: 40), context token limits, text_only parsing mode, and user_prompt for response formatting. This file is automatically created with defaults when a KB is created and can be viewed/edited using the config CLI command.

1. Knowledge Base Directory Structure: When you create knowledge bases, they are stored within the directory specified by knowledge_base.base_dir in your main config.yaml. The structure typically looks like this:

    <base_dir>/              # Main directory, contains a set of knowledge bases
    ├── config.yaml          # Main application configuration (copied from config.example.yaml)
    ├── .env                 # Environment variables referenced in config.yaml
    ├── kbmcp.log
    ├── knowledge_base_1/    # Directory for the first KB
    │   ├── config.yaml      # KB-specific configuration (query parameters)
    │   ├── <storage_files>  # The LightRAG storage files
    └── knowledge_base_2/    # Directory for the second KB
        ├── config.yaml
        ├── <storage_files>

5. New Features

5.1 Text-Only Document Parsing

By default, knowledge-mcp processes documents using both text content and metadata (like document structure, formatting, etc.). You can now configure knowledge bases to use text-only parsing for faster processing and reduced token usage.

Benefits:

• Faster document processing
• Lower LLM token consumption
• Simplified content extraction
• Better performance with large document collections

Configuration: Add text_only: true to your knowledge base's config.yaml:

# In <base_dir>/<kb_name>/config.yaml
description: "My knowledge base with text-only parsing"
mode: "hybrid"
top_k: 40
text_only: true  # Enable text-only parsing

Usage:

# Create a new KB and configure it for text-only parsing
knowledge-mcp --config config.yaml create my_text_kb
knowledge-mcp --config config.yaml config my_text_kb edit
# Add text_only: true to the config file

# Add documents - they will be processed with text-only parsing
knowledge-mcp --config config.yaml add my_text_kb ./documents/

5.2 Configurable User Prompts

You can now customize how the LLM formats and structures its responses for each knowledge base by configuring a user_prompt. This allows you to tailor the response style to match your specific use case.

Benefits:

• Consistent response formatting across queries
• Domain-specific response styles
• Better integration with downstream applications
• Improved user experience

Configuration: Add a user_prompt field to your knowledge base's config.yaml. The prompt supports multi-line YAML syntax:

# In <base_dir>/<kb_name>/config.yaml
description: "Technical documentation KB"
mode: "hybrid"
top_k: 30
max_entity_tokens: 2000
max_relation_tokens: 3000
max_total_tokens: 6000
response_type: "Multiple Paragraphs"
user_prompt: |
  Please provide a comprehensive analysis with citations.
  - Focus on actionable information
  
  Keep your response clear, concise, and well-organized.

Example Configurations:

1. Business-Focused Format:

user_prompt: |
  Structure your response for business stakeholders:
  
  **Executive Summary** (2-3 sentences)
  Brief overview of the main points and business impact.
  
  **Key Findings**
  • Most critical insights
  • Relevant metrics or data points
  • Risk factors or opportunities
  
  **Recommendations**
  • Specific actionable steps
  • Priority levels (High/Medium/Low)
  • Expected outcomes

1. Technical Documentation Style:

user_prompt: |
  You are a technical documentation expert. Please structure your response with:
  
  1. **Context**: Brief background on the topic
  2. **Implementation**: Step-by-step technical details
  3. **Best Practices**: Recommended approaches and common pitfalls
  4. **Examples**: Concrete code examples or use cases where applicable
  
  Use clear headings, bullet points, and code blocks for readability.

1. Academic Research Style:

user_prompt: |
  Please provide a scholarly response that includes:
  
  • **Introduction**: Context and scope of the topic
  • **Analysis**: Critical examination of key concepts and evidence
  • **Synthesis**: How different pieces of information connect
  • **Conclusion**: Main findings and implications
  
  Support your points with specific references from the knowledge base.

Usage:

# Configure user prompt for an existing KB
knowledge-mcp --config config.yaml config my_kb edit
# Add your user_prompt configuration to the YAML file

# Query the KB - responses will follow your configured format
knowledge-mcp --config config.yaml query my_kb "What are the main concepts?"

Notes:

• User prompts are applied automatically to all queries for that knowledge base
• Leave user_prompt empty or omit it to use default LLM behavior
• Changes take effect immediately - no need to rebuild the knowledge base
• Backward compatible - existing knowledge bases continue to work without modification

6. Usage (CLI)

The primary way to interact with knowledge-mcp is through its CLI, accessed via the knowledge-mcp command (if installed globally or via uvx knowledge-mcp within the activated venv).

All commands require either `--config <path-to-config.yaml>` or `--base <directory-containing-config.yaml>` (e.g. Docker uses `--base /app/kb`).

uv run knowledge-mcp --config /path/to/config.yaml shell

Available Commands (Interactive Shell):

Example (Direct CLI):

# Create a knowledge base named 'my_docs'
knowledge-mcp --config config.yaml create my_docs

# Add a document to it
knowledge-mcp --config config.yaml add my_docs ./path/to/mydocument.pdf

# Search the knowledge base
knowledge-mcp --config config.yaml query my_docs "What is the main topic?"

# Start the interactive shell
knowledge-mcp --config config.yaml shell

(kbmcp) list
(kbmcp) query my_docs "Another query"
(kbmcp) exit

7. Development

1. Project Decisions

• Tech Stack: Python 3.12, uv (dependency management), hatchling (build system), pytest (testing).
• Setup: Follow the installation steps, ensuring you install with uv pip install -e ".[dev]".
• Code Style: Adheres to PEP 8.
• Testing: Run tests using uvx test or pytest.
• Dependencies: Managed in pyproject.toml. Use uv pip install <package> to add and uv pip uninstall <package> to remove dependencies, updating pyproject.toml accordingly.
• Scripts: Common tasks might be defined under [project.scripts] in pyproject.toml.
• Release: Build hatch build and then twine upload dist/*.

1. Test with uvx

    "knowledge-mcp": {
      "command": "uvx",
      "args": [
        "--project",
        "/path/to/knowledge-mcp",
        "knowledge-mcp",
        "--config",
        "/path/to/knowledge-mcp/kbs/config.yaml",
        "mcp"
      ]
    }

1. Test with MCP Inspector

npx @modelcontextprotocol/inspector uv "run knowledge-mcp --config /path/to/config.yaml mcp"

or

npx @modelcontextprotocol/inspector uvx --project . knowledge-mcp "--config ./kbs/config.yaml mcp

1. Convenience dev scripts

Assumes a local config file at ./kbs/config.yaml

• uvx shell - Starts the interactive shell
• uvx insp - Starts the MCP Inspector