Skip to content

custom_mcp_tool.md

Latest commit

 

History

History
78 lines (64 loc) · 2.54 KB

custom_mcp_tool.md

File metadata and controls

78 lines (64 loc) · 2.54 KB

Define Your Custom MCP Tool

MiroRL supports two main approaches for adding custom MCP tools:

1. Using npx/uvx (Node.js-based MCP tools)

For Node.js-based MCP tools that can be installed via npm and executed with npx or uvx, add your tool configuration file:

tools:
  - class_name: "mirorl.tools.mcp_tool.MCPTool"
    config:
      command: "npx"  # or "uvx"
      args: ["your-mcp-tool-name"]
      env: ["YOUR_API_KEY", "HTTPS_PROXY"]  # Required environment variables
      server_name: "your_tool_name"
      max_retries: 5
      delay_between_retries: 1
      connection_timeout: 5
      execution_timeout: 60
    tool_schema:
      type: "mcp"
      function: {}

2. Using Python-based MCP tools

For Python-based MCP tools using the FastMCP framework, add your tool configuration file:

tools:
  - class_name: "mirorl.tools.mcp_tool.MCPTool"
    config:
      command: "python"
      args: ["path/to/your_tool.py"]
      env: ["YOUR_API_KEY", "HTTPS_PROXY"]  # Required environment variables
      server_name: "your_tool_name"
      max_retries: 5
      delay_between_retries: 1
      connection_timeout: 5
      execution_timeout: 60
    tool_schema:
      type: "mcp"
      function: {}

3. Configuration Parameters

All MCP tools support the following configuration parameters:

Parameter Description Default Example
command MCP server command to execute - "npx", "python"
args Command line arguments [] ["your-tool"], ["tool.py"]
env Environment variables [] ["API_KEY", "PROXY"]
server_name MCP server identifier - "your_tool_name"
blacklist Forbidden operations [] ["scrape"]
max_retries Maximum retry attempts 5 3
delay_between_retries Retry delay in seconds 1 2
connection_timeout Connection timeout in seconds 5 10
execution_timeout Execution timeout in seconds 60 120

4. Integration with Training

To use your custom MCP tool in training:

  1. Create your tool configuration file in mirorl/recipe/mcp/config/tool_config/
  2. Set environment variables required by your tool
  3. Update training command to use your tool config:
python3 -m mirorl.recipe.mcp.main_ppo \
    --config-path="mirorl/recipe/mcp/config" \
    --config-name='mcp_trainer' \
    actor_rollout_ref.rollout.multi_turn.tool_config_path="path/to/your_tool_config.yaml" \
    data.train_batch_size=256 \
    trainer.n_gpus_per_node=8