// Agent profile
BrowserSkill (Tencent/BrowserSkill) is an open-source tool that bridges AI agents and your actual browser session. Instead of spinning up headless browsers or requiring separate credentials, BrowserSkill lets any shell-capable AI agent — Claude Code, Aider, custom scripts — perform browser automation through your real, logged-in Chrome instance. It ships as a CLI plus a companion browser extension. The CLI exposes browser actions (navigate, click, type, read page, take screenshots) as tool calls that agents can invoke, while the extension handles the connection to your active tabs without hijacking focus or disrupting your workflow. MIT licensed, TypeScript-based, works across macOS and Linux.
BrowserSkill solves a practical problem in AI agent tooling: agents that need to interact with authenticated web applications usually require either headless browser setups with duplicated credentials or brittle screen-scraping approaches. BrowserSkill takes a different path — it lets agents use your actual browser session, the one where you’re already logged into GitHub, Jira, your company dashboard, and everything else.
The architecture is straightforward. A browser extension runs in your Chrome instance and exposes page interaction capabilities over a local connection. A CLI tool provides the agent-facing interface — standard tool calls for navigation, clicking, typing, reading page content, and taking screenshots. The agent calls the CLI, the CLI talks to the extension, the extension acts on the page. Your browser stays usable throughout.
Real session access: The core value proposition is that agents inherit your existing browser sessions. No OAuth token management, no cookie copying, no separate login flows. If you can see it in your browser, your agent can interact with it.
Agent-agnostic design: BrowserSkill works with any agent that can invoke shell commands. It ships with integration examples for popular frameworks but imposes no framework lock-in. The CLI accepts structured input and returns structured output — any agent that can call a tool can use it.
Non-disruptive automation: Unlike tools that take over your browser window, BrowserSkill can operate on background tabs while you continue working in your foreground tab. The extension manages the coordination to avoid focus theft.
Developers use BrowserSkill to give their coding agents access to documentation sites, issue trackers, and CI dashboards without manual copy-paste. Teams building custom agents use it as the browser interaction layer rather than maintaining their own Playwright or Puppeteer setups. It is particularly useful for internal tools and enterprise applications where headless browser authentication is complex or impossible.
BrowserSkill requires Chrome and the companion extension installed. It operates on your real browser session, which means the agent has the same access you do — appropriate trust boundaries and agent sandboxing are important. Currently supports Chrome only; Firefox support is on the roadmap. The tool is relatively new (5K+ stars) but growing quickly.
BrowserSkill is built for developers and agent builders who need their AI tools to interact with authenticated web applications. It appeals to teams building internal agent workflows, developers who want browser automation without headless complexity, and anyone tired of copying content between their browser and their agent’s terminal.
Builder.io's open-source framework for building agent-native applications — shared actions, SQL-backed state, identity, tools, skills, jobs, observability, and UI surfaces that all work together.
CLI orchestration tool that installs and configures platform connectors for 16+ social and web sources — Twitter/X, Reddit, YouTube, GitHub, Bilibili, and more. Zero API fees.
Persistent memory layer for AI coding agents — benchmark-backed (95.2% on LongMemEval-S), 92% fewer tokens per session vs full-context pasting, zero manual memory.add() calls.