Add MCP tool calling to any OpenAI-compatible LLM via a stateless fetch proxy.
{ "mcpServers": { "mcp-completions-stateless": { "command": "npx", "args": [ "-y", "mcp-completions-stateless" ] } } }