Heurist Mesh MCP Server

Overview

Heurist Mesh provides a framework for creating, deploying, and using specialized AI agents. These agents can be accessed through both synchronous and asynchronous APIs, as well as through the Model Context Protocol (MCP) which allows AI assistants like Claude to directly interact with Mesh agents as tools.

Installation

To use Heurist Mesh via MCP, you can either:

1. Use the Heurist Mesh MCP Portal - The fastest way to get started with no setup required
2. Self-host using the heurist-mesh-mcp-server

To add the Heurist Mesh MCP server to your configuration, include the following in your MCP configuration:

{
  "mcpServers": {
    "heurist-mesh": {
      "url": "https://mcp.heurist.ai"
    }
  }
}

Authentication

To use any Mesh agent, you'll need a Heurist API key. You can obtain one at https://heurist.ai/credits.

Using Mesh Agents

Mesh agents can be accessed through two interfaces:

1. Synchronous API: Provides direct, immediate responses for quick queries and actions
2. Asynchronous API: Designed for longer-running tasks or when you want to track the reasoning process

For detailed API documentation including examples of both synchronous and asynchronous usage, refer to the official Heurist Mesh documentation.

Building Custom Mesh Agents

If you want to create your own Mesh agent:

1. Set up your development environment:

cd mesh
uv sync
source .venv/bin/activate  # Linux/Mac
.venv\Scripts\activate     # Windows

2. Create a new agent by extending the MeshAgent class:

from mesh.mesh_agent import MeshAgent
from typing import Dict, Any, List

class MySpecialAgent(MeshAgent):
    def __init__(self):
        super().__init__()
        self.metadata.update({
            'name': 'My Special Agent',
            'version': '1.0.0',
            'author': 'Your Name',
            'author_address': '0xYourEthereumAddress',
            'description': 'This agent can do...',
            'external_apis': ['API_Name'],
            'tags': ['Category1', 'Category2'],
            'image_url': 'https://example.com/image.png',
            'examples': ['Example query 1', 'Example query 2'],
        })

    def get_system_prompt(self) -> str:
        """Return the system prompt for the agent"""
        return """
        You are a helpful assistant that can [describe agent's purpose].
        [Include any specific instructions for the LLM here]
        """

    def get_tool_schemas(self) -> List[Dict]:
        """Define the tools that your agent exposes"""
        return [
            {
                "type": "function",
                "function": {
                    "name": "my_tool_name",
                    "description": "Description of what this tool does",
                    "parameters": {
                        "type": "object",
                        "properties": {
                            "param1": {"type": "string", "description": "Description of parameter 1"},
                            "param2": {"type": "number", "description": "Description of parameter 2"},
                        },
                        "required": ["param1"]
                    }
                }
            }
        ]

3. Implement the necessary methods for your agent's functionality

4. Test your agent locally before submitting it to the Mesh network

Once a Mesh agent is added to the Github main branch, it's automatically deployed and instantly available via REST API and MCP.