What problem does it solve?
This Skill provides reliable, repeatable guidance for running interactive coding agents (Codex, Claude Code, OpenCode, Pi) in terminal environments so they do not hang, lose formatting, or read unintended files during programmatic workflows.
Core Features & Use Cases
- PTY-first execution: Emphasizes allocating a pseudo-terminal (pty:true) so interactive CLIs render correctly and remain responsive.
- Background session orchestration: Start agents in background mode, receive a sessionId, and use process actions to list, poll, retrieve logs, send input, and kill sessions.
- Safe workspace handling: Use workdir and temporary git repos or worktrees for reviews and batch PR fixes to avoid editing live Moltbot or sensitive directories.
- Notification and monitoring: Attach wake triggers so Moltbot is notified on completion and follow the progress update rules to keep users informed.
Quick Start
Start a background Codex agent in the project workdir with pty:true and monitor its logs until it reports completion.