BrowserSkill BrowserSkill
BrowserSkill connects shell-capable AI coding agents (Cursor, Claude Code, Codex, OpenClaw, CodeBuddy, WorkBuddy, Pi, Hermes Agent, DeepSeek Harness) to the Chromium browser you are already logged into. It has two local pieces: the `bsk` CLI/daemon and a Chrome/Edge extension. Browser tasks run in a separate, visible Agent Window so you can keep working, agents reuse your real login state without test accounts, and touching one of your own tabs requires an explicit borrow-and-return. When a task hits a captcha, login, or confirmation dialog the agent hands control to you and resumes afterwards. `bsk install-skill` installs a harness-specific skill that teaches the agent to open pages, read text, fill forms, scrape data, run UI regression checks on a PR, and capture full-page screenshots (`bsk screenshot --full-page`). A sandboxed-agent mode keeps the daemon in a persistent host environment via a shared `BSK_HOME`.
cli-v0.3.0 released 2026-09-17; ~5k GitHub stars and trending on GitHub (daily, +1.3k stars) in mid-September 2026 after adding DeepSeek Harness support and sandboxed-agent setup docs.
As of
- Agents drive your real, logged-in Chrome or Edge without separate test accounts
- Isolated visible Agent Window; your own tabs are protected unless explicitly borrowed
- Built-in human-in-the-loop handoff for captchas, logins, and confirmations
- Harness-agnostic `bsk` CLI; `bsk install-skill` for Cursor, Claude Code, Codex, OpenClaw, Pi, Hermes
- Full-page screenshots, form filling, scraping, and UI regression checks
- macOS, Linux, and Windows; sandboxed-agent daemon mode
Developers who want their coding agent to verify deployed pages, test UI flows, or pull data from sites they are signed into, without handing over credentials or interrupting their own browsing.
Headless CI browser automation (use Playwright or a hosted browser API) or Firefox users (support is planned). For a desktop app alternative see Superagent; for agent-driven E2E testing see Kane CLI.
Free and open-source under MIT; install the `bsk` CLI via the install script (macOS/Linux) or PowerShell (Windows) and the extension from the Chrome Web Store or Edge Add-ons. Runs entirely locally with your existing agent and model.
A clean, safety-conscious way to give any coding agent a real browser; the explicit tab-borrowing and human handoff design set it apart from raw CDP bridges.
Is BrowserSkill free?
Yes - BrowserSkill is Open Source. Free and open-source under MIT; install the `bsk` CLI via the install script (macOS/Linux) or PowerShell (Windows) and the extension from the Chrome Web Store or Edge Add-ons. Runs entirely locally with your existing agent and model.
Is BrowserSkill open source?
Yes - BrowserSkill is open source. Free and open-source under MIT; install the `bsk` CLI via the install script (macOS/Linux) or PowerShell (Windows) and the extension from the Chrome Web Store or Edge Add-ons. Runs entirely locally with your existing agent and model.
Who is BrowserSkill best for?
Developers who want their coding agent to verify deployed pages, test UI flows, or pull data from sites they are signed into, without handing over credentials or interrupting their own browsing.
Who is BrowserSkill not ideal for?
Headless CI browser automation (use Playwright or a hosted browser API) or Firefox users (support is planned). For a desktop app alternative see Superagent; for agent-driven E2E testing see Kane CLI.
What are the best BrowserSkill alternatives?
The closest BrowserSkill alternatives on ai.dosa.dev are Pieces for Developers, agenttrace, Vibe Kanban - all listed under Developer Productivity & Workflow.
Who makes BrowserSkill?
BrowserSkill is developed by Tencent. It is listed in the Developer Productivity & Workflow category on ai.dosa.dev.
Favorite this tool to revisit it later, or Zap it to contribute to the public vote count.
Content on this page is AI-generated. Please verify details with the vendor's website for accuracy.