# Agentic Resource Registry (ARR)

> Agentic Resource Registry (ARR, formerly Observal) is a self-hosted, open source (Apache-2.0) registry for the resources AI coding agents use: MCP servers, skills, hooks, prompts and sandboxes, bundled into agents. Teams publish and review them, then pull approved agents into Claude Code, Codex CLI, Cursor, Kiro, GitHub Copilot, Copilot CLI, OpenCode, Goose, Pi or Antigravity with one command. ARR also traces every session through hooks and writes Insights reports on each agent.

Key facts:

- Runs on your own infrastructure with Docker Compose (PostgreSQL, ClickHouse, Redis); sessions and traces never leave it.
- The CLI is `observal` (install with `pipx install observal-cli`); `observal agent pull <team>/<agent> --harness <tool>` installs an agent into a coding tool.
- Teamspaces give each team its own namespace; components and agents go through review before others can install them.
- Speaks Agentic Resource Discovery (ARD), so ARD clients can ask the registry what exists for a task.
- Insights: reports per agent, from its real sessions, naming what keeps going wrong, the numbers behind it, and prompt changes that fix it.

## Links

- [Website](https://agenticresourceregistry.io/): product overview and FAQ
- [Book a demo](https://agenticresourceregistry.io/book-a-demo): talk to the team
- [Source code](https://github.com/AgenticResourceRegistry/AgenticResourceRegistry): GitHub repository, README and releases
- [Documentation](https://docs.observal.io/): setup, CLI reference and guides
- [Blog](https://agenticresourceregistry.io/blog): guides on coding agent observability, cost and evaluation

## Guides

- [Setting Up a Self-Hosted AI Observability Platform in 10 Minutes](https://agenticresourceregistry.io/blog/self-hosted-ai-observability-platform-setup)
- [What Is AI Agent Observability and Why Your Team Needs It](https://agenticresourceregistry.io/blog/what-is-ai-agent-observability)
- [How to Evaluate AI Coding Agents: A Practical Scoring Framework](https://agenticresourceregistry.io/blog/ai-agent-evaluation-scoring-guide)
- [How to Reduce AI Agent Costs by 40% Without Losing Productivity](https://agenticresourceregistry.io/blog/reduce-ai-agent-costs-engineering-teams)
- [Claude Code vs Cursor: Which AI Coding Agent Performs Better in 2026?](https://agenticresourceregistry.io/blog/claude-code-vs-cursor-performance-comparison)