# BrowserSkill

> Updated 2026-09-16 · type: tool · category: mcp-agents · status: active · rev 1

BrowserSkill lets an AI agent borrow a live tab from your logged-in browser and act with your real sessions, instead of driving a blank headless browser.

- Open source: yes (MIT)
- Self-hostable: yes
- Pricing model: free
- Best for: Anyone pointing an agent at logged-in web apps — a RevOps person who wants an agent to pull a report out of a dashboard that has no API, or a researcher who needs it to read pages behind a login — without standing up separate auth or a scraping proxy.
- Last verified: 2026-09-16

- **Canonical:** https://gtmstacker.com/registry/tool/browserskill/
- **Source:** [github · tencent/browserskill](https://github.com/tencent/browserskill)
- **Tags:** mcp-agents, browser-automation, agent-tooling, self-hostable, session-reuse
- **Repository:** https://github.com/tencent/browserskill

## Is BrowserSkill open source?

Yes, BrowserSkill is open source under the MIT license.

## How much does BrowserSkill cost?

BrowserSkill is free to use.

## Can I self-host BrowserSkill?

Yes, BrowserSkill can be self-hosted (the source is available under the MIT license).

## Alternatives & related

- [DeepScrape](https://gtmstacker.com/registry/tool/deepscrape/)


---

MIT-licensed Rust tool from Tencent that bridges an AI agent to the real browser you are already signed into, rather than a fresh headless one. A local CLI plus a browser extension let the agent borrow one of your open tabs, drive it using the cookies and sessions you already have, and then return control to you. The effect is that the agent operates inside your existing auth state, so it can reach logged-in pages without a scripted login or credentials handed to a bot.

## Provenance

- MIT, Rust, ~2.7k stars and ~500 commits independently WebFetch-verified on the repo 2026-09-16 (github.com/tencent/browserskill, published under the tencent org). The borrow-a-tab / hand-it-back model and the local daemon + extension design are from the repo README.
- Surfaced via the 2026-09-16 viral-posts brief as the highest-signal drop of the pass (the Tencent announcement post, ~1.5k likes / ~130k views inside 8 hours).
- Curated from the GTM Stacker signal registry (2026-09-16 pass: daily pull + viral-posts brief); license independently verified 2026-09-16.

## Why it matters for a GTM stack

Most agent-browser tools give the model a blank browser and then you spend the afternoon teaching it to log in. A huge amount of the GTM data an agent actually needs lives behind a login with no clean API — a CRM report view, an ad-account dashboard, a partner portal. BrowserSkill's move is to let the agent ride your existing authenticated session in a tab you choose, which removes the login-automation problem entirely. The same property is the risk: an agent on a borrowed tab can do anything you can do on that site, so this is a lend-one-tab-deliberately tool, not something to point at your whole browser. Used with that discipline, it is the shortest path from "the data is stuck in a logged-in dashboard" to "an agent can read it today."