LoopTroop

LoopTroop is an open-source, local GUI application that orchestrates AI coding agents to complete multi-file tickets. Unlike chat assistants or simple CLI tools, it is built to manage complex, long-running tasks across multiple repositories without filling the model's context window with irrelevant history.

How it works:

1. Planning: Multiple language models act as a council. They draft plans, vote anonymously, and merge the best approaches.
2. Context management: The application stores files, specs, and intermediate states outside the model. It builds a fresh, minimal context window for each specific phase, which reduces model confusion.
3. Task isolation: The tool splits the main ticket into small tasks called beads. Each task executes in its own git worktree.
4. Error recovery: If a task fails or times out, the app logs the specific error, resets the git worktree, and retries with a fresh context that contains only the failure note.

Who it is for: LoopTroop is for developers who want a local, visual dashboard to monitor and review AI-generated code changes. It requires a local installation and uses the models configured in your OpenCode setup. It does not replace simple IDE autocompletion; instead, it is designed for larger, multi-file code changes that require structured execution and review before merging.