kevo-mcp

KevoDB/kevo-mcp

3.2

If you are the rightful owner of kevo-mcp and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

The KevoDB MCP Server allows AI agents to interact with KevoDB using a standardized API through the Multimodal Communication Protocol.

Tools

  1. connect

    Connect to the KevoDB server

  2. get

    Get a value by key from KevoDB

  3. put

    Store a key-value pair in KevoDB

  4. delete

    Delete a key-value pair from KevoDB

  5. scan

    Scan keys in KevoDB with options

  6. batch_write

    Perform multiple operations in a batch

  7. get_stats

    Get database statistics

  8. begin_transaction

    Begin a new transaction and return transaction ID

  9. commit_transaction

    Commit a transaction by ID

  10. rollback_transaction

    Roll back a transaction by ID

  11. tx_put

    Store a key-value pair within a transaction

  12. tx_get

    Get a value by key within a transaction

  13. tx_delete

    Delete a key-value pair within a transaction

  14. cleanup

    Close the KevoDB connection

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

  1. Install dependencies:
pip install fastmcp python-kevo
  1. Ensure KevoDB is running on localhost:50051 (or set the KEVO_HOST and KEVO_PORT environment 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:

ToolDescription
connectConnect to the KevoDB server
getGet a value by key from KevoDB
putStore a key-value pair in KevoDB
deleteDelete a key-value pair from KevoDB
scanScan keys in KevoDB with options
batch_writePerform multiple operations in a batch
get_statsGet database statistics
begin_transactionBegin a new transaction and return transaction ID
commit_transactionCommit a transaction by ID
rollback_transactionRoll back a transaction by ID
tx_putStore a key-value pair within a transaction
tx_getGet a value by key within a transaction
tx_deleteDelete a key-value pair within a transaction
cleanupClose the KevoDB connection

Integration with AI Applications

To use KevoDB with your AI application:

  1. Start the KevoDB server
  2. Start this MCP server
  3. Configure your AI agent to connect to the MCP endpoint
  4. The AI agent can now use all KevoDB operations through the MCP interface

License

MIT