القائمة
Control Electron desktop applications (Slack, Discord, Notion, VS Code) using Vercel's agent-browser via Chrome DevTools Protocol (CDP).
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
| name | agent-browser |
| description | Control Electron desktop applications (Slack, Discord, Notion, VS Code) using Vercel's agent-browser via Chrome DevTools Protocol (CDP). |
| version | 1 |
| author | Antigravity |
[!IMPORTANT] This skill MUST be executed strictly under the Omni-Architect Agent Protocol v1.0. All tool executions, code modifications, and communications MUST adhere to the 13 core protocols.
Version: 1.0.0 Status: ACTIVE Category: 10-tools
This skill utilizes Vercel Labs' agent-browser CLI to allow autonomous AI agents to interact with and control desktop applications built on Electron (Slack, Discord, Notion, Figma, VS Code, Spotify) or WebView2. It circumvents the need for complex APIs or OAuth, allowing the agent to "see" and "click" the UI directly via the Chrome DevTools Protocol (CDP).
To use this tool, the AI Agent must execute the following CLI command:
npx skills add vercel-labs/agent-browser --skill electron
Instead of passing the entire bloated HTML of a desktop application (which consumes massive context limits), agent-browser extracts a Snapshot + Refs representation. This is a clean, compressed snapshot of only the interactive elements on the screen.
To control an Electron app, it must be launched with remote debugging enabled.
The user must launch their Electron app from the terminal with the --remote-debugging-port flag.
Examples:
# MacOS Examples
/Applications/Notion.app/Contents/MacOS/Notion --remote-debugging-port=9222
/Applications/Slack.app/Contents/MacOS/Slack --remote-debugging-port=9222
/Applications/Discord.app/Contents/MacOS/Discord --remote-debugging-port=9222
Once the port (e.g., 9222) is active, instruct the agent to connect via the --cdp flag using agent-browser.
# Using the CLI to connect to the active CDP port
agent-browser --cdp 9222
# Or full WebSocket URL if remote
agent-browser --cdp http://localhost:9222
When tasking an agent with agent-browser, use explicit instructions:
"Use
agent-browserconnected to CDP port9222. Read the latest messages in the active Slack channel, summarize them, and type a response clicking the 'Send' button."
Consolidated master domain for agentic arch.
Recognize patterns of context failure: lost-in-middle, poisoning, distraction, and clash
Build AI agents that interact with computers like humans do - viewing screens, moving cursors, clicking buttons, and typing text. Covers Anthropic's Computer Use, OpenAI's Operator/CUA, and open-source alternatives. Critical focus on sandboxing, security, and handling the unique challenges of vision-based control. Use when: computer use, desktop automation agent, screen control AI, vision-based agent, GUI automation.
Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking. Use when testing LLM performance, measuring AI application quality, or establishing evaluation frameworks.
Production-ready patterns for building LLM applications. Covers RAG pipelines, agent architectures, prompt IDEs, and LLMOps monitoring. Use when designing AI applications, implementing RAG, building agents, or setting up LLM observability.
Design short-term, long-term, and graph-based memory architectures