What problem does it solve?
Running and reproducing terminal-based agent behaviors is error-prone and environment-sensitive; this Skill provides an isolated, repeatable tmux-based workflow to run OpenConductor, reproduce user-reported issues, and validate attention-detection heuristics without affecting a user's real configuration.
Core Features & Use Cases
- Isolated E2E Testing: Launch OpenConductor in a detached tmux session with isolated config, logs, and state to reproduce bugs safely.
- Terminal Capture & Visuals: Capture rendered terminal text and produce pixel-accurate screenshots for debugging and Telegram notifications.
- Interactive Input Simulation: Send keystrokes and mouse events into the session to reproduce permission dialogs, question dialogs, long-running spinners, and scrollback behaviors.
- Log Inspection & Validation: Tail structured JSON logs to verify attention checks, auto-approve behavior, and Telegram bridge events.
- Use Case: Reproduce a user bug where a Claude Code spinner is missed, verify attention classifier output, and capture the terminal snapshot to attach to an issue.
Quick Start
Build the OpenConductor binary, start it in a detached tmux session with isolated config and logs, then capture the visible screen and logs for analysis.