Orbi
Autonomous software delivery runner that turns labelled GitHub Issues into implemented, independently reviewed pull requests, merges and tagged releases on your own infrastructure
Quick Start
git clone https://github.com/orbi-build/orbi.git && cd orbi && uv tool install --force --reinstall --editable --python 3.14 . Overview
Orbi is a self hosted autonomous software delivery runner built around GitHub Issues. Add the ai-ready label to an issue and Orbi creates an isolated worktree, asks an AI coding agent to implement and test the change, opens a pull request, runs a separate review session and merges the reviewed commit. It can then create a tagged GitHub release.
GitHub provides the task queue and permanent record, so Orbi does not need its own database or project dashboard. A scheduled runner checks for work through systemd on Linux or launchd on macOS. Installation requires Python 3.14, a recent GitHub CLI, Pi and access to an OpenAI compatible model provider. Container images are also available.
The self hosted runner is free under AGPL 3.0, with an alternative Sustainable Use License. A paid hosted service is available for teams that do not want to operate the runner. Your repository and credentials stay on your machine when self hosting, although prompts and relevant source context go to the model provider you configure.
Orbi receives substantial repository access. It can push branches, modify workflow files, merge pull requests and publish releases. A Git worktree separates changes but does not provide a security sandbox. Run it on isolated infrastructure with narrowly scoped GitHub permissions. Its independent review is another AI session rather than human approval, and macOS support has not yet received the same real hardware verification as Linux. The project is young and changes quickly, so pinning releases is prudent.
Use Cases
Specific ways to use Orbi for your workflow.
Deployment Strategy
Recommended ways to host Orbi in your own environment.