Observal is a self-hosted control plane and registry for internal AI components, providing server+CLI deployment, an agent registry, and analytics features for Skills, MCPs, and Agents.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Observal is the control plane and system of record for internal AI components.
How it works
- It offers a registry to browse, search, and install agents with harness compatibility badges.
- It treats agents as portable context packages containing MCP servers, skills, hooks, prompts, and sandboxes, and generates per-harness config files.
- It provides insights and session replay to analyze usage and debugging data.
Getting started
Two parts: a server to self-host (API + web UI + databases) and a CLI on developer machines.
Deploy the server
curl -fsSL https://raw.githubusercontent.com/Observal/Observal/main/install-server.sh | bash
Install the CLI
Standalone binary (no Python required):
curl -fsSL https://raw.githubusercontent.com/Observal/Observal/main/install.sh | bash
Python (3.11+):
uv tool install observal-cli
# or: pipx install observal-cli
Connect your harness
observal auth login
observal doctor --patch
Recent releases
- v1.11.0 Observal v1.11.0 (2026-08-02)
- v1.10.7 Observal v1.10.7 (2026-07-27)
- v1.10.6 Observal v1.10.6 (2026-07-23)
- v1.10.5 Observal v1.10.5 (2026-07-22)
- v1.10.4 Observal v1.10.4 (2026-07-22)
Notable changes
- Added features such as allowing dots in namespaces (registry) and improved linting/audit workflows.
Traction
2254 stars, 462 forks, 242 open issues.
Behind the repo
The project provides a server+CLI architecture with deployment docs, self-hosting overview, and production deployment guidance.
Caveats
License: Apache-2.0. Language: Python. Created 2026-03-30. Last push 2026-08-03.