Introduction

What is OneMCP?

OneMCP is a production-ready runtime that lets LLMs operate tools safely and autonomously. It combines a knowledge base, retrieval, and dynamic code generation to translate goals into executable actions, while emitting rich telemetry so you can observe, debug, and improve behavior. It’s model-agnostic and designed to plug into your stack via the Model Context Protocol.

The Core Problem it Solves

Designing and wiring tools is tedious: you must define schemas, bind APIs, manage auth, keep prompts in sync with code, and add logging/tracing that’s often an afterthought. OneMCP reduces this toil by indexing your documentation and APIs, generating the glue code on demand, and standardizing observability—so developers spend less time hand-crafting integrations and more time on outcomes.

Key Features

• Knowledge-driven: indexes Agent.md, docs, and OpenAPI to ground tool use with retrieval.
• Dynamic code generation: synthesizes safe, executable calls and workflows from instructions.
• Built-in observability: OpenTelemetry traces/metrics/logs for every step and tool call.
• Pluggable models: works with OpenAI, Gemini, Anthropic, and more via a simple provider interface.

Get Started in 5 Minutes

curl -sSL https://raw.githubusercontent.com/Gentoro-OneMCP/onemcp/main/cli/install.sh | bash
onemcp chat