Forge Orchestrator by Nxtg AI is a free MCP server tool designed for developers managing multiple AI clients on shared repositories. Built in Rust and distributed as a single, dependency-free binary, it coordinates diverse AI activities like those from Claude Code, Codex CLI, and Gemini CLI. The program uses a .forge/ filesystem for state management, operating without a daemon or traditional database. Its core function is to prevent conflicts, capture knowledge, and manage task execution across various AI tools working on the same codebase, enhancing development workflows.

What does Forge Orchestrator connect to?

Forge Orchestrator functions as an MCP server, primarily connecting to various AI clients such as Claude Code, Codex CLI, and Gemini CLI. It coordinates their operations on shared code repositories. Its internal state management relies entirely on the .forge/ filesystem, avoiding the need for a daemon or a conventional database. The orchestrator uses forge/ filesystem and MCP stdio as its transport protocols to communicate with these connected AI clients. This setup allows it to expose its full orchestration state to the connected clients.

Core features and tools

The orchestrator provides several callable command-line tools for interaction. These include forge init for project initialization, forge config brain openai for configuring AI brains, and forge plan --generate for task generation. Users can view the task board with forge status, execute tasks headlessly using forge run, or launch an interactive TUI dashboard via forge dashboard --pty. Specific tasks can be run with designated agents using commands like forge run --task T-001 --agent codex.

Who is it for?

Forge Orchestrator is designed for developers who are actively running multiple AI tools on the same codebase. Its purpose is to help prevent conflicts, ensure knowledge capture, and efficiently manage task execution in these complex environments. It supports cohesive operations by providing features like file locking and drift detection. Skip this if your development workflow does not involve coordinating multiple AI tools on a single shared repository.

How to get started

The server authenticates using an API key passed as an environment variable. For instance, connecting to an OpenAI brain requires an OPENAI_API_KEY.

Use cases

• Multi-AI task orchestration on shared repositories, offering file locking, knowledge capture, task planning, and drift detection.
• Structured development from specification to execution, involving project initialization, defining requirements in SPEC.md or README, and generating dependency-aware task graphs.
• Autonomous task execution for CI/CD pipelines or headless operations, alongside interactive management through a terminal user interface dashboard for real-time multi-agent orchestration.

Forge Orchestrator offers a free, Rust-based solution for coordinating multiple AI clients on shared codebases. It manages state via the filesystem and provides command-line tools for task planning, execution, and monitoring. The system is ideal for developers needing to unify diverse AI operations and maintain project integrity.