takt

Why we included this project

Most AI coding agents work fine until the task gets long, then they start forgetting instructions or blurring steps together. TAKT fixes that by making the workflow explicit. You define planning, implementation, review, and fix loops in YAML, and each step runs with its own persona, permissions, and output contract, so the agent only ever sees the context it needs. Tasks run in isolated worktrees, and every step leaves logs and reports, which means you can trace exactly how a pull request came together. It works with Claude Code, Codex, OpenCode, Cursor, and GitHub Copilot CLI, so teams already using those tools can adopt it without switching providers. The same orchestration model also handles non-coding multi-agent tasks that need review and feedback loops.