Project Orchestrator

Project Orchestrator coordinates AI coding agents for developers working across tasks or related repositories. It gives agents a shared record of code, plans and decisions so each task can draw on earlier work.

• Shared code context: Agents can search code by meaning and inspect its structure and relationships, with automatic updates as files change.
• Plans and execution: Task plans track dependencies and progress, and an autonomous runner dispatches work to agents in parallel.
• Reusable workflows: Code reviews and release pipelines can follow repeatable workflows, including nested workflows with tracked execution states.
• Architectural decisions: Proposals move through review, acceptance and implementation, with notes linked to affected code and plans.
• Related projects: Workspaces group repositories around shared context, contracts and milestones, while skills can move between projects and instances.

MCP connects the shared context to Claude Code, OpenAI Agents and Cursor. Code parsing supports Rust, TypeScript, Python, Go, Swift and other languages.

The coordination backend is self-hosted and requires Neo4j and Meilisearch. macOS and Linux installation, desktop apps and Docker deployment are available. Optional NATS integration synchronizes events between instances; without it, MCP changes don't propagate as live events to the rest of the system. The conversational interface connects to Claude.