agentregistry
Fast-track AI innovation with a centralized, trusted, curated registry
claude mcp add --transport stdio agentregistry-dev-agentregistry docker run -i agentregistry/agentregistry:latest \ --env AR_HOST="http://localhost:12121 (default registry UI/API host)" \ --env AR_PORT="12121 (default registry UI/API port)"
How to use
Agent Registry provides a centralized, governance-focused registry for AI artifacts, including MCP servers, agents, and skills. The MCP ecosystem enables you to publish, discover, and deploy AI artifacts securely across environments. With the Agent Registry, you can manage artifact lifecycles, validate data, and streamline deployment workflows through a single registry interface and the Agent Gateway, which exposes MCP endpoints to IDEs and automation tools. The primary tools involved are the arctl CLI and the web UI, which together allow you to list MCP servers, seed data, and interact with published artifacts.
To use the MCP server, install and run the Docker-based image (as shown in the installation steps) or opt for the provided CLI to manage the registry. Once running, you can use arctl mcp list to discover available MCP servers, start the registry daemon if needed, and explore built‑in seed data. You can also configure IDEs via arctl to integrate Claude Desktop, Cursor, or VS Code configurations so your AI IDEs can access the MCP endpoints via the Agent Gateway, enabling read_file, create_issue, and other MCP-provided tools through a single gateway URL.
How to install
Prerequisites:
- Docker Desktop with Docker Compose v2+ (for containerized usage)
- Optional: Go 1.25+ if you plan to build from source (as per prerequisites in the repo)
- Optional: arctl CLI if you want to run the registry locally outside of the Docker container
Installation steps (Docker-based):
- Pull and run the Agent Registry Docker image:
docker run -d --name ar-registry -p 12121:12121 agentregistry/agentregistry:latest
- Verify the service is up and listening on port 12121 (web UI/API):
curl -sS http://localhost:12121/health || echo "not healthy"
- If you prefer to use the arctl CLI for MCP management, install it:
# Install via script (recommended)
curl -fsSL https://raw.githubusercontent.com/agentregistry-dev/agentregistry/main/scripts/get-arctl | bash
- Start interacting with MCP servers via arctl:
arctl mcp list
Prerequisites for building from source (Go-based server):
- Go 1.25+ installed
- Git to clone the repository
- Docker for building and testing containers (optional but recommended)
Build from source (if you clone locally):
git clone https://github.com/agentregistry-dev/agentregistry.git
cd agentregistry
go build ./...
Additional notes
Notes and tips:
- The official UI is served at http://localhost:12121 by default. Adjust port mappings if you change the container settings.
- The Agent Gateway provides a single MCP endpoint for all deployed servers, enabling clients to discover and invoke tools uniformly.
- If you encounter connectivity issues, ensure Docker networking allows localhost access and that the arctl CLI and container image versions are compatible.
- The registry focuses on governance and secure curation of artifacts; consider enabling data enrichment and scoring pipelines as outlined in the repository to improve artifact quality signals.
- Environment variables such as AR_HOST and AR_PORT can be used to override defaults for scripted deployments or CI/CD environments.
Related MCP Servers
mindsdb
Query Engine for AI Analytics: Build self-reasoning agents across all your live data
adk-go
An open-source, code-first Go toolkit for building, evaluating, and deploying sophisticated AI agents with flexibility and control.
kagent
Cloud Native Agentic AI | Discord: https://bit.ly/kagentdiscord
context-space
Ultimate Context Engineering Infrastructure, starting from MCPs and Integrations
station
Station is our open-source runtime that lets teams deploy agents on their own infrastructure with full control.
agents
AI agent tooling for data engineering workflows.