{
  "id": 8307549,
  "title": "BrowserSkill: Connecting AI Coding Agents to Your Logged-In Browser",
  "url": "https://urgent.news/2026/09/18/browserskill-connecting-ai-coding-agents-to-your-logged-in-browser",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-18T20:48:09.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/terminalchai/browserskill-connecting-ai-coding-agents-to-your-logged-in-browser-3c9c"
  },
  "original_language": "en",
  "account": "Tencent has released BrowserSkill, an open-source tool that connects AI coding agents directly to a user's logged-in browser. This allows agents to access authenticated web applications, staging dashboards, and other web services without needing to create separate test accounts or API tokens. BrowserSkill achieves this by securely bridging the agent to the developer's existing browser profile, inheriting all session cookies, local storage, and active logins automatically.\n\nThe tool operates on a three-tier architecture: the AI coding agent (like Claude Code or Cursor), the `bsk` CLI and daemon (written in Rust), and the Chromium extension for Chrome and Microsoft Edge. The daemon handles session orchestration and IPC messaging, while the extension listens to the daemon and controls browser tabs through native extension APIs. An agent-controlled agent window is spawned to perform tasks without interrupting the user's primary browsing session.\n\nBrowserSkill offers several key capabilities:\n1. Reuse of real login state, allowing agents to interact with authenticated web apps and services.\n2. Human-in-the-loop assistance for tasks that require user intervention, such as solving CAPTCHAs or confirming security dialogs.\n3. Tab borrowing protocol, enabling agents to inspect or modify tabs opened by the user, with user approval and automatic tab return after completion.\n4. Full-page high-resolution screenshots for multimodal debugging.\n\nSetting up BrowserSkill is straightforward with an automated installation script for Claude Code, Cursor, Codex, or any shell-capable agent. Once installed, users can verify connectivity, start sessions, and open pages to inspect their functionality. The tool also integrates with DeepSeek Harness, providing native browser_* tool definitions and live previews within the Harness UI.",
  "summary": "One of the biggest hurdles in modern AI-assisted software development is giving agents reliable access to the web. Whether you are using Claude Code, Cursor, Codex, or custom autonomous agents, software engineering rarely happens in a vacuum. Developers constantly need agents to verify UI flows on localhost, check staging dashboards, inspect documentation, or automate repetitive web tasks. Until…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}