gwbischof/outsource-mcp

Overview

A Python-based MCP server that lets an AI assistant delegate subtasks to other AI assistants. It works by spawning child agents that can run independently and report back results.

Best for

Best for
Developers building multi-agent systems who want to experiment with hierarchical delegation in MCP

Use cases

• Offloading complex multi-step reasoning to parallel sub-agents
• Running independent research or data gathering tasks concurrently
• Building hierarchical agent workflows where a primary agent coordinates helpers

How to use

Tools exposed

• uvx

Tested with

Claude Desktop, Cline, ChatGPT

Example client config

{\n  "mcpServers": {\n    "outsource-mcp": {\n      "command": "uvx",\n      "args": ["--from", "git+https://github.com/gwbischof/outsource-mcp.git", "outsource-mcp"],\n      "env": {\n        "OPENAI_API_KEY": "your-openai-key",\n        "ANTHROPIC_API_KEY": "your-anthropic-key",\n        "GOOGLE_API_KEY": "your-google-key",\n        "GROQ_API_KEY": "your-groq-key",\n        "DEEPSEEK_API_KEY": "your-deepseek-key",\n        "XAI_API_KEY": "your-xai-key",\n        "PERPLEXITY_API_KEY": "your-perplexity-key",\n        "COHERE_API_KEY": "your-cohere-key",\n        "FIREWORKS_API_KEY": "your-fireworks-key",\n        "HUGGINGFACE_API_KEY": "your-huggingface-key",\n        "MISTRAL_API_KEY": "your-mistral-key",\n        "NVIDIA_API_KEY": "your-nvidia-key",\n        "OLLAMA_HOST": "http://localhost:11434",\n        "OPENROUTER_API_KEY": "your-openrouter-key",\n        "TOGETHER_API_KEY": "your-together-key",\n        "CEREBRAS_API_KEY": "your-cerebras-key",\n        "DEEPINFRA_API_KEY": "your-deepinfra-key",\n        "SAMBANOVA_API_KEY": "your-sambanova-key"\n      }\n    }\n  }\n}

Notes

A Python-based MCP server that lets an AI assistant delegate subtasks to other AI assistants. It works by spawning child agents that can run independently and report back results.

28 stars on GitHub. Last updated 2025-05-28. Licensed MIT.

Use cases

• Offloading complex multi-step reasoning to parallel sub-agents
• Running independent research or data gathering tasks concurrently
• Building hierarchical agent workflows where a primary agent coordinates helpers

Pros

• Enables parallel task execution for faster completion
• Simple Python implementation with minimal dependencies
• Extends MCP protocol to support agent delegation

Cons

• Small community and limited documentation (28 stars)
• Requires managing multiple API keys and costs for each sub-agent
• No built-in error handling or retry logic for failed sub-tasks

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.