Connects Claude to Harness's continuous delivery and cloud cost management platform through their API. The server promises comprehensive API coverage with enterprise features like rate limiting, retry logic, and async support, though the docs don't spell out specific operations like triggering pipelines, managing deployments, or querying cost data. You'll need Harness API credentials in your environment to get it running. Reach for this if you're using Harness for CI/CD or FinOps and want to query deployment status, manage infrastructure configs, or pull cost analytics without leaving your AI workflow. Written in Python with stdio transport.
A Model Context Protocol (MCP) server for integrating Harness with GenAI applications.
Continuous delivery and cloud cost management
pip install harness-mcp-server
Or install from source:
git clone https://github.com/asklokesh/harness-mcp-server.git
cd harness-mcp-server
pip install -e .
Create a .env file or set environment variables according to Harness API requirements.
from harness_mcp import HarnessMCPServer
# Initialize the server
server = HarnessMCPServer()
# Start the server
server.start()
MIT License - see LICENSE file for details
HARNESS_API_KEY*secretAPI Key for HARNESS
HARNESS_API_URLAPI URL for HARNESS