A complete enterprise-grade MCP (Model Context Protocol) platform featuring server hosting, a unified registry, a secure gateway, and a chat client with built-in RAG, supporting both Docker and Kubernetes deployments.
Obot is an open-source project maintained by Obot AI (formerly Acorn Labs) designed to address the governance challenges of connecting AI agents to internal enterprise systems. Built on the MCP (Model Context Protocol) standard, it provides a complete platform comprising Hosting, Registry, Gateway, and Chat Client components.
Core Components
- MCP Hosting: Supports Docker for local runs or Kubernetes for production. Compatible with Node.js, Python, and containerized MCP servers, supporting both STDIO (single-user) and HTTP (multi-user) modes.
- MCP Registry: Offers MCP-compliant server discovery and catalog management. The platform handles shared credentials centrally and controls server visibility based on user permissions.
- MCP Gateway: Acts as a unified entry point providing request routing, content review, audit logging, and granular access control rules.
- Obot Chat: A chat client with built-in RAG capabilities, supporting multiple model providers like OpenAI and Anthropic. It features project-wide memory, reusable configurations, and scheduled task automation.
Technical Implementation
The stack uses Go (backend) and Svelte/TypeScript (frontend), with PostgreSQL 17+ (with pgvector extension) for the data layer. Security features include built-in OAuth 2.1 and Token authentication, at-rest/in-transit encryption, and comprehensive audit logs. It supports GitOps for configuration management. It is suitable for enterprises building private MCP catalogs or providing standardized tool invocation interfaces for IDE plugins (Cursor, VSCode) or automation platforms.