nerve-bridge-skill_skill

Overview

This skill provides a bi-directional bridge to the Trae IDE on macOS using AppleScript and a file-based feedback loop. It sends instructions into Trae, waits for a confirmation file, and returns JSON-formatted completion feedback. Use it to run commands or code inside Trae and reliably detect when the work is finished.

How this skill works

The skill copies your instruction to the system clipboard and uses osascript to activate Trae and paste the payload into its input. Instructions must include a small Python hook that writes a JSON signal file to ~/.openclaw/workspace/trae_feedback.json when the task completes. The bridge polls for that file for up to 300 seconds and returns the parsed JSON when detected.

When to use it

• Execute code or commands inside Trae and require deterministic completion confirmation.
• Automate multi-step edits or test runs in Trae from external scripts on macOS.
• Build workflows that need closed-loop acknowledgement from the IDE before proceeding.

Best practices

• Always append the required Python hook that writes the feedback JSON to ~/.openclaw/workspace/trae_feedback.json.
• Keep the feedback payload small and include a status field and timestamp for reliability.
• Ensure Terminal and System Events have Accessibility permissions on macOS.
• Activate and focus the Trae window before sending input to avoid paste failures.
• Account for the 5-minute timeout by keeping tasks short or emitting interim signals if needed.

Example use cases

• Create a new file in Trae, add code, and wait for a confirmation that tests ran successfully.
• Trigger a build or lint operation inside Trae and proceed only after the feedback file reports success.
• Automated refactors where each change must be validated by a Trae-side script before continuing.