sub-agent MCP Server

A Model Context Protocol (MCP) server that allows sending chat completion messages to any OpenAI-compatible endpoint, with support for background job tracking using goroutines and in-memory storage with TTL.

Features

• Send chat completion requests to OpenAI-compatible endpoints
• Background job tracking with asynchronous execution
• In-memory storage with configurable TTL for results
• Three MCP tools for managing sub-agent calls

Configuration

The server is configured via environment variables:

Variable	Description	Default
OPENAI_BASE_URL	The base URL for the OpenAI API endpoint	https://api.openai.com/v1
OPENAI_MODEL	The model to use for chat completions	gpt-3.5-turbo
OPENAI_API_KEY	The API key for authentication	Required
TTL	Time-to-live for stored results (Go duration format)	1h

MCP Tools

sub_agent_send

Send a chat completion message to an OpenAI-compatible endpoint.

Input:

• message (string, required): The message to send to the OpenAI endpoint
• background (boolean, optional): Whether to run the request in the background (default: false)
• model (string, optional): Override the default model for this request

Output:

• If synchronous: Returns the completion result directly
• If background: Returns a task ID for tracking

sub_agent_list

List all active sub-agent calls with their status and creation time.

Input: None

Output:

• results: Array of sub-agent result objects
• count: Number of active results

sub_agent_get_result

Get the result of a completed sub-agent call by task ID.

Input:

• task_id (string, required): The task ID to retrieve the result for

Output:

• result: The sub-agent result object, or error if not found/expired

Usage

Building

cd sub-agent
go build -o sub-agent main.go

Running

export OPENAI_BASE_URL="https://your-openai-compatible-endpoint/v1"
export OPENAI_MODEL="your-model"
export OPENAI_API_KEY="your-api-key"
export TTL="2h"

./sub-agent

Example: Synchronous Request

{
  "tool": "sub_agent_send",
  "arguments": {
    "message": "Hello, how are you?"
  }
}

Example: Background Request

{
  "tool": "sub_agent_send",
  "arguments": {
    "message": "Generate a summary of the following text...",
    "background": true
  }
}
// Returns: {"task_id": "uuid", "status": "queued"}

Example: List All Jobs

{
  "tool": "sub_agent_list",
  "arguments": {}
}

Example: Get Result

{
  "tool": "sub_agent_get_result",
  "arguments": {
    "task_id": "uuid-from-background-request"
  }
}

Architecture

• In-memory Store: Uses a thread-safe map with read/write locks for concurrent access
• TTL Cleanup: Background goroutine periodically removes expired entries (runs every minute)
• OpenAI Client: Uses the github.com/sashabaranov/go-openai library for API calls
• UUID Generation: Uses github.com/google/uuid for unique task IDs

Dependencies

• github.com/modelcontextprotocol/go-sdk: MCP protocol implementation
• github.com/sashabaranov/go-openai: OpenAI API client
• github.com/google/uuid: UUID generation