Skynet-MCP (THIS PROJECT IS A WORK IN PROGRESS)

What is Skynet-MCP?

Skynet-MCP is a hierarchical network of AI agents using the Model Context Protocol (MCP). Each instance acts as both an MCP server and client, allowing agents to spawn child agents with shared tools to decompose and parallelize complex tasks.

How to use Skynet-MCP?

To use Skynet-MCP, clone the repository, install dependencies (Node.js v20+, npm, and API keys for OpenAI or Anthropic), then run npm start or use Docker Compose via npm run docker:up. The server provides MCP tools Invoke and DelayedResponse for creating and managing agent tasks.

Key features of Skynet-MCP

• Dual-mode operation as both MCP server and client
• Hierarchical agent management with child agent spawning
• Integration with OpenAI or Anthropic LLMs
• Automatic tool discovery from connected MCP servers
• Multiple transport options: STDIO and SSE
• Delayed execution for asynchronous task polling

Use cases of Skynet-MCP

• Decompose complex research tasks into parallel sub‑tasks handled by child agents
• Automate multi‑step coding workflows by spawning specialized code agents
• Create reporting pipelines where agents gather, analyze, and summarize data
• Build scalable AI services that distribute work across a network of models

FAQ from Skynet-MCP

How does Skynet-MCP differ from a single MCP server?

Skynet-MCP creates a recursive agent network where each agent can spawn child agents with the same toolset, enabling parallel task decomposition rather than sequential processing.

What runtime and dependencies are required?

Node.js v20+ and npm. Optional Docker and Docker Compose for containerized development. API keys for OpenAI or Anthropic are needed for LLM‑powered agents.

Where does task state and data live?

State can be stored in memory or Redis, configurable via the PERSISTENCE_TYPE environment variable and REDIS_URL.

What transports and authentication are supported?

STDIO and SSE transport are supported. Authentication is not detailed; the README focuses on environment‑variable configuration for API keys.

Is this project production‑ready?

The README explicitly marks it as "A WORK IN PROGRESS" and notes the agent framework still needs implementation. It is not yet production‑ready.