Agentic Orchestrator
An open-source AI coding orchestrator that carries feature requests through planning, implementation, review and pull request publication.
Agentic Orchestrator coordinates coding agents for engineers supervising feature work across one or more repositories. It keeps requirements, design decisions and verification evidence attached to the work as it moves toward a reviewable pull request.
- Repository context: Build a reusable knowledge base of architecture, conventions, dependencies and verification methods, then carry research and design findings into implementation.
- Phased development: Break a feature into a roadmap and detailed phase plans, with shorter workflows for well-understood changes and deeper review for complex or high-risk work.
- AI review gates: Review plans for architecture and scope, add security, performance and testing checks where appropriate, and run a final review and fix loop before publication.
- Human checkpoints: Optional approval gates pause for questions or reviews of the research, design and plans; manual publication lets you inspect the diff and pull request description first.
- Parallel and cross-repository work: Give each feature separate Git worktrees and branches, order work by repository dependencies and link related pull requests.
One provider can handle the whole workflow, or multiple providers can divide the work. Automatic publication can commit changes, push branches and create pull requests. The desktop app also supports holding completed work for manual review before publication.
The project is open source under Apache 2.0. Worktrees remain separate from your main checkout, and their cleanup is manual.
