pgalyen1987/gate402-mcp

Overview

MCP server for Gate402, a pay-per-call agent API over x402/USDC. It exposes 20 tools for renting GPU/CPU compute per second, running LLM inference on open models (Llama 3.1, Qwen, Mistral), executing a full Base trading loop, and scraping web pages to Markdown. Includes a free tier and requires no signup.

Best for

Best for
Agent builders who want a single MCP bridge to paid compute, inference, trading, and web scraping.

Use cases

• Rent per-second GPU or CPU compute from an agent
• Call Llama 3.1, Qwen, or Mistral inference on demand
• Run the Base trading loop from token discovery to execution

Notes

MCP server for Gate402, a pay-per-call agent API over x402/USDC. It exposes 20 tools for renting GPU/CPU compute per second, running LLM inference on open models (Llama 3.1, Qwen, Mistral), executing a full Base trading loop, and scraping web pages to Markdown. Includes a free tier and requires no signup.

2 stars on GitHub. Last updated 2026-08-03. Licensed MIT.

Use cases

• Rent per-second GPU or CPU compute from an agent
• Call Llama 3.1, Qwen, or Mistral inference on demand
• Run the Base trading loop from token discovery to execution

Pros

• 20 tools in a single MCP server, covering compute, inference, trading, and scraping
• Free tier and no signup lower the barrier to trying it
• Pay-per-call and per-second billing aligns cost with actual usage

Cons

• Very low community traction (2 stars) makes long-term maintenance uncertain
• Depends on x402/USDC settlement and Gate402’s specific API behavior
• JavaScript MCP server limits integration to runtimes with Node support

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