launch

Automate VS Code and Code OSS UI interactions via Chrome DevTools Protocol.

1|Updated Apr 8, 2026
One-click install
npx skills add https://github.com/voidful/Aixlarity --skill launch-voidful
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: launch
Source: https://github.com/voidful/Aixlarity/tree/main/aixlarity-ide/.agents/skills/launch
Command: npx skills add https://github.com/voidful/Aixlarity --skill launch-voidful

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires agent-browser.

What problem does it solve? Manually testing VS Code UI features, chat panel behavior, and extension workflows is slow and hard to verify. This Skill lets an agent launch VS Code (Code OSS) with remote debugging and drive the UI programmatically through agent-browser, with screenshots as evidence. ## Core Features & Use Cases - CDP-Based Automation: Launch Code OSS, the Agents workbench, or extension debug sessions with --remote-debugging-port and connect agent-browser to snapshot and interact with UI elements. - Monaco Editor Input Handling: Reliable patterns for typing into the chat input across builds, including the keyboard-shortcut focus plus per-key press fallback when type @ref silently fails on Code OSS. - Screenshot Evidence: Timestamped screenshot folders capture UI state after launch, before and after interactions, and on failures. - Use Case: After modifying VS Code chat panel source code, rebuild, relaunch Code OSS on port 9224, focus the chat input, type a prompt, send it, and screenshot the result to verify the change. ## Quick Start Launch Code OSS with remote debugging on port 9224, connect agent-browser, snapshot the interactive elements, and take a screenshot of the chat panel.

Frequently Asked Questions about launch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate VS Code UI interactions from the command line?

Launch VS Code or Code OSS with --remote-debugging-port, then connect agent-browser to that CDP port. Use agent-browser snapshot -i to discover interactive elements and interact with them via element refs or keyboard press commands.

How to type text into the VS Code chat input programmatically?

Focus the chat input with the keyboard shortcut (Ctrl+Cmd+I on macOS, Ctrl+Alt+I on Linux/Windows), then send each character with agent-browser press commands. This works on all builds, while type @ref silently fails on Code OSS.

Why does agent-browser type command not work in Code OSS?

The type @ref, keyboard type, and keyboard inserttext commands silently fail on Code OSS because Monaco Editor does not process those input events there. Use the keyboard shortcut to focus the input, then send individual press commands per key.

Can I debug a VS Code extension using Chrome DevTools Protocol?

Yes. Compile the extension, then launch VS Code Insiders with --extensionDevelopmentPath, --remote-debugging-port, and a separate --user-data-dir. Connect agent-browser to the port and snapshot the UI to interact with the extension host window.

Why does VS Code say Sent env to running instance and exit?

VS Code detects an already-running instance and forwards your arguments to it, so the new process exits and CDP never starts. Quit the existing instance or pass a unique --user-data-dir to force a separate process with remote debugging.

How do I take screenshots of VS Code during automation?

Run agent-browser screenshot with a file path inside a timestamped directory. On macOS, your terminal needs Screen Recording permission; otherwise use eval-based DOM checks to verify UI state without screenshots.