KevoDB MCP Server
This project implements a MCP (Multimodal Communication Protocol) server for KevoDB, allowing AI agents to interact with KevoDB using a standardized API.
Features
- Exposes KevoDB operations through MCP tools
- Supports all core KevoDB functionality:
- Basic key-value operations (get, put, delete)
- Range, prefix, and suffix scans
- Transactions
- Batch operations
- Database statistics
- Simple string-based API with UTF-8 encoding
Prerequisites
- Python 3.8+
- Running KevoDB server (default: localhost:50051)
- FastMCP library
- Python-Kevo SDK
Installation
- Install dependencies:
pip install fastmcp python-kevo
- Ensure KevoDB is running on localhost:50051 (or set the
KEVO_HOSTandKEVO_PORTenvironment variables to connect to a different endpoint)
Usage
Running the MCP Server
Start the MCP server:
python main.py
This will launch the MCP server on http://localhost:9000/mcp
You can configure the KevoDB connection using environment variables:
KEVO_HOST: Hostname of the KevoDB server (default: "localhost")KEVO_PORT: Port of the KevoDB server (default: "50051")
Example:
KEVO_HOST=192.168.1.100 KEVO_PORT=5000 python main.py
Using with AI Agents
AI agents that support MCP can connect to this server and use all exposed tools. The server provides the following tools:
| Tool | Description |
|---|---|
connect | Connect to the KevoDB server |
get | Get a value by key from KevoDB |
put | Store a key-value pair in KevoDB |
delete | Delete a key-value pair from KevoDB |
scan | Scan keys in KevoDB with options |
batch_write | Perform multiple operations in a batch |
get_stats | Get database statistics |
begin_transaction | Begin a new transaction and return transaction ID |
commit_transaction | Commit a transaction by ID |
rollback_transaction | Roll back a transaction by ID |
tx_put | Store a key-value pair within a transaction |
tx_get | Get a value by key within a transaction |
tx_delete | Delete a key-value pair within a transaction |
cleanup | Close the KevoDB connection |
Integration with AI Applications
To use KevoDB with your AI application:
- Start the KevoDB server
- Start this MCP server
- Configure your AI agent to connect to the MCP endpoint
- The AI agent can now use all KevoDB operations through the MCP interface