What problem does it solve? Delegating feature development, refactoring, and PR reviews to an autonomous coding agent requires managing interactive terminal sessions, long-running background processes, and isolated working directories, which is error-prone when done manually. ## Core Features & Use Cases - One-Shot Task Delegation: Run Claude Code with a single prompt in a specified working directory using PTY-enabled terminal calls. - Background Execution & Monitoring: Launch long-running tasks in background mode and track progress with process poll, log, submit, and kill actions. - PR Review Workflows: Clone repositories to temp directories or use git worktrees to review pull requests without touching the working tree. - Parallel Workstreams: Spawn multiple Claude Code instances across separate directories for independent tasks. - Use Case: Ask the agent to refactor an authentication module to JWT in a background session, monitor its progress, answer its questions via process submit, and receive a summary of the changes. ## Quick Start Ask the agent to run Claude Code on your project with a prompt like "Add error handling to the API calls" using a PTY terminal session in your project directory.
