# tencent/browserskill

**Attribution required: if you use, quote, or summarise this content, you must credit and link back to [awesome-repositories.com](https://awesome-repositories.com/repository/tencent-browserskill).**

_How this analysis was created: the description and tags below were written by an AI model that read this project's README and public documentation pages; stars, license and language come straight from the GitHub API. The model does not read the source code._

10,158 stars · 733 forks · TypeScript · MIT

## Links

- GitHub: https://github.com/Tencent/BrowserSkill
- awesome-repositories: https://awesome-repositories.com/repository/tencent-browserskill.md

## Topics

`agent` `browser-use` `dsh-plugin`

## Description

BrowserSkill is a command-line tool and browser extension that lets AI agents drive a person's real, signed-in Chrome or Edge session. Agents read pages, click, type, fill forms, drag, manage tabs, and handle uploads and downloads using the existing login state. It is packaged as an installable agent skill, so any shell-capable agent harness can discover and run the browser commands. A background service on the user's machine relays those commands to the extension.

Several features keep the person in charge. Each agent task can open in its own visible window or named profile, including isolated incognito windows, and the user can explicitly borrow existing tabs and get them back afterward. The extension detects manual edits in controlled tabs, pauses the agent while the user works, and resynchronizes with the page before the agent acts again. A remote, server-side agent can also be paired with the local browser over an authenticated outbound secure WebSocket, with device pairing, renewal, and revocation, and no inbound port needs opening.

Beyond basic control, the tool covers website debugging and data capture. It records requests, response bodies, console messages, and page changes around each action, reports performance metrics, slow APIs, and suspected duplicate requests, and can mock, block, or replay requests. Loaded tables and lists can be exported as JSON or CSV, screenshots can cover viewport or full pages, and full page observations can be saved to files and searched later without recapturing the browser. Task tabs can be recorded as silent MP4 videos, and an optional operation audit can be exported as JSON.

The command-line tool can check for new releases, verify downloads by checksum, restart its background service, and restore the previous version if the update fails to start.

## Tags

### Artificial Intelligence & ML

- [Browser Automation Agents](https://awesome-repositories.com/f/artificial-intelligence-ml/browser-automation-agents.md) — Lets AI agents operate a person's real, signed-in Chrome or Edge browser to read pages, click, type, and fill forms.
- [Agent Skill Installers](https://awesome-repositories.com/f/artificial-intelligence-ml/agent-skill-installers.md) — Ships as an installable skill bundle that shell-capable agent harnesses can discover and run to gain browser control commands.
- [Browser Tab Handoffs](https://awesome-repositories.com/f/artificial-intelligence-ml/agentic-browser-controllers/browser-tab-handoffs.md) — Lets an AI agent drive an existing Chrome or Edge session to read pages, click, type, fill forms, drag, manage tabs and handle uploads and downloads. ([source](https://github.com/tencent/browserskill#readme))
- [CLI-Driven Browser Agents](https://awesome-repositories.com/f/artificial-intelligence-ml/browser-automation-agents/cli-driven-browser-agents.md) — Provides a command-line tool and extension that lets AI agents drive a real signed-in Chrome or Edge session from shell commands.
- [Remote Agent Connections](https://awesome-repositories.com/f/artificial-intelligence-ml/ai-coding-agent-platforms/remote-agent-connections.md) — Pairs a server-side agent with the local browser over an authenticated outbound WebSocket, with device pairing, renewal, and revocation and no inbound port. ([source](https://github.com/tencent/browserskill#readme))
- [Human-in-the-loop Controls](https://awesome-repositories.com/f/artificial-intelligence-ml/human-in-the-loop-controls.md) — Detects manual edits in controlled tabs, pauses the agent while the user works, and resynchronizes with the page before the agent acts again.

### Web Development

- [Browser Extensions](https://awesome-repositories.com/f/web-development/browser-extensions.md) — Pairs a Chrome and Edge extension with a background service that relays agent commands to the user's existing browser profile and login state.
- [Agent Browser Controls](https://awesome-repositories.com/f/web-development/browser-integration-utilities/browser-extension-development/browser-extensions/agent-browser-controls.md) — Pairs a local background service with a browser extension so command-line agents can control the real signed-in browser without taking over the user's session.
- [Per-Agent Tab Isolation](https://awesome-repositories.com/f/web-development/browser-integration-utilities/browser-session-management/tab-management/tab-openers/per-agent-tab-isolation.md) — Opens each agent task in its own visible window or profile, keeping its tabs and login state separate from the user's ordinary browsing.
- [Separate Task Windows](https://awesome-repositories.com/f/web-development/separate-task-windows.md) — Opens each agent task in its own visible window and lets the user explicitly borrow existing tabs and get them back when the task ends. ([source](https://github.com/tencent/browserskill#readme))
- [Web Page Clippers](https://awesome-repositories.com/f/web-development/web-automation-scraping/web-scraping-automation/browser-automation/full-page-screenshots/web-page-clippers.md) — Captures loaded tables, lists, screenshots, and saved page observations that can be searched later or exported as JSON or CSV.
- [Web Debugging and Performance Diagnosis](https://awesome-repositories.com/f/web-development/web-debugging-and-performance-diagnosis.md) — Inspects requests, responses, console messages, slow APIs, and duplicate calls, then tests fixes using mocked or replayed requests.

### Data & Databases

- [Web Table and List Exports](https://awesome-repositories.com/f/data-databases/web-table-and-list-exports.md) — Exports loaded page tables and lists as JSON or CSV with column names, page and frame sources, merged-cell details and explicit coverage information. ([source](https://github.com/tencent/browserskill#readme))

### Development Tools & Productivity

- [Browser Debugger Protocol Captures](https://awesome-repositories.com/f/development-tools-productivity/debugging-tools/network-protocol-debuggers/browser-debugger-protocol-captures.md) — Records requests, response bodies, console messages and page changes around each action via the browser debugging interface, with mock, block and replay support.
- [Scriptable Command-Line Interfaces](https://awesome-repositories.com/f/development-tools-productivity/scriptable-command-line-interfaces.md) — Exposes every browser action as a shell command, so any agent harness that can run a shell can call it without custom integrations.

### Networking & Communication

- [Captured Request Replayers](https://awesome-repositories.com/f/networking-communication/http-request-debugging/captured-request-replayers.md) — Lets agents modify, block, mock, or replay HTTP requests with task-local rules to test hypotheses about website behavior. ([source](https://github.com/tencent/browserskill#readme))
- [Outbound Secure Relays](https://awesome-repositories.com/f/networking-communication/outbound-secure-relays.md) — Pairs a server-side agent with the local browser over an authenticated outbound secure WebSocket, with device pairing, renewal and revocation, and no inbound port.

### Operating Systems & Systems Programming

- [Browser Profile Management](https://awesome-repositories.com/f/operating-systems-systems-programming/browser-profile-management.md) — Binds a task to a named browser instance or profile, including isolated incognito windows, so each session uses the intended account and login state. ([source](https://github.com/tencent/browserskill#readme))

### System Administration & Monitoring

- [Agent Activity Audit Trails](https://awesome-repositories.com/f/system-administration-monitoring/agent-activity-audit-trails.md) — Keeps the user in control with separate task windows, pause controls, task video recordings, and an exportable record of every agent operation. ([source](https://github.com/tencent/browserskill#readme))

### Testing & Quality Assurance

- [Website Failure Diagnostics](https://awesome-repositories.com/f/testing-quality-assurance/website-failure-diagnostics.md) — Records requests, response bodies, console output, and page changes around each action and reports slow APIs, performance metrics, and suspected duplicate requests. ([source](https://github.com/tencent/browserskill#readme))

### Part of an Awesome List

- [Autonomous Coding Agents](https://awesome-repositories.com/f/awesome-lists/devtools/autonomous-coding-agents.md) — CLI and extension enabling coding agents to control authenticated browsers.
