AgentCraft

AgentCraft MCP Server enables secure communication between your MCP client and intelligent AgentCraft AI Agents. It supports premade and custom agents, providing a scalable, enterprise-ready integration for exchanging data and commands with agents.

How to use

You run the AgentCraft MCP Server locally or on your infrastructure and connect to it from your MCP client. The server exposes endpoints and tooling that let you send data to an agent, receive responses from agents, and retrieve the list of available agents. Use the dedicated MCP client integration to exchange messages and trigger agent actions, while keeping authentication and environment configuration under your control.

How to install

Prerequisites you need before installing this server:

• Python 3.8+ must be installed on the system where you run the MCP server.
• You can verify with python --version.

Step 1: Install the MCP server package using Python's package manager.

pip install agentcraft-mcp

Step 2: Run the MCP server to start listening for MCP client connections.

python -m agentcraft_mcp.server

Step 3: Configure your MCP client to connect to the AgentCraft MCP Server. You will typically provide the server as an endpoint or a local stdio interface, depending on your deployment. The server supports a local stdio runtime setup as shown in the configuration example below.

Configuration example for Claude Desktop integration is provided here so you can wire the MCP server into your client configuration.

Additional sections

Configuration details for connecting to the AgentCraft MCP Server are shown in the following snippet. It defines a named MCP server called AgentCraft that runs via Python and uses the module path agentcraft_mcp.server.

{
  "mcpServers": {
      "AgentCraft": {
          "command": "python",
          "args": [
              "-m",
              "agentcraft_mcp.server"
          ],
          "env": {
              "AGENTCRAFT_BEARER_TOKEN": "Your AgentCraft Bearer Token for authorization",
              "ENVIRONMENT": "PRODUCTION"
          }
      }
  }
}

Available tools

send_agent_data

Send data to an agent for processing or actioning based on your prompt.

receive_agent_data

Receive data or responses from an agent using your query and tracking key.

get_available_agents

Query and list all available agents that can be used by the MCP client.