This project provides a centralized, curated registry for AI artifacts, including MCP servers, agents, skills, and prompts, aiming to accelerate AI innovation. It targets organizations and developers by offering a trusted, single source for AI building blocks, simplifying discovery, management, and deployment, thereby reducing manual configuration overhead and improving developer onboarding.

How It Works

AgentRegistry acts as a central catalog for AI components, accessible via a CLI (arctl) and a web UI. It allows users to import or publish artifacts from diverse sources like npm, PyPI, Docker Hub, and GitHub repositories. The platform facilitates one-click deployment and automatic IDE configuration. It integrates tightly with agentgateway, an AI-native reverse proxy, which provides a unified, secure entry point for AI clients to access deployed MCP servers and agents, offering features like authentication, centralized observability, and dynamic discovery.

Quick Start & Requirements

• Installation: Install the CLI using curl -fsSL https://raw.githubusercontent.com/agentregistry-dev/agentregistry/main/scripts/get-arctl | bash.
• Prerequisites: Docker Desktop with Docker Compose v2+.
• Running: Start the agentregistry daemon by running any arctl command (e.g., arctl version). The UI is accessible at http://localhost:12121.
• Semantic Search: Requires a PostgreSQL instance with the pgvector extension enabled.
• Links: Website, Docs, Discord, GitHub, Quick Start (URLs not provided in README).

Highlighted Details

• Supports registration of MCP servers from npm, PyPI, OCI/Docker images, and remote HTTP/SSE endpoints.
• Enables building and publishing skills (knowledge packages), agents (identity + dependencies), and prompts (instruction templates).
• agentgateway offers a single endpoint for AI clients, JWT validation, on-behalf-of auth flows, and OTEL-compatible observability.
• Semantic search for artifacts is powered by pgvector.
• IDE integration for Claude Desktop, Cursor, and VS Code is supported via agentgateway.

Maintenance & Community

Community interaction is encouraged via a Discord channel. Establishing community meetings is on the project's roadmap. Contribution guidelines are detailed in CONTRIBUTING.md, with architecture and development setup in DEVELOPMENT.md.

Licensing & Compatibility

The project is licensed under the Apache 2.0 license, generally permissive for commercial use and integration with closed-source projects.

Limitations & Caveats

Community meetings are not yet established. Semantic search functionality requires explicit configuration of PostgreSQL with the pgvector extension, which is disabled by default.