weaveq/onmyagent/browser-skill

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Drive the user's real Chromium browser (Chrome/Edge logins and cookies) through the external Tencent BrowserSkill `bsk` CLI and extension. Use for logged-in sites, form flows, page reads, and human-in-the-loop captcha/login handoff. Prefer purpose-built MCP connectors or OnMyAgent in-app Browser (browser-automation) when they fit better. Requires `bsk` on PATH and the BrowserSkill extension connected (green popup). If setup is missing, tell the user to open Extensions → BrowserSkill (guided 3-step panel).

Install

npx vskill@latest install weaveq/onmyagent/browser-skill

SKILL.md

BrowserSkill (external real browser)

Use the external BrowserSkill stack (bsk CLI + Chrome/Edge extension). This is not OnMyAgent's in-app browser (browser-automation) and not Computer Use (desktop AX).

Choose the right surface

SurfaceWhat it isPrefer when
browser-automationOnMyAgent in-app browser (right rail Browser panel)Web work without needing the user's real Chrome/Edge logins; in-app preview
browser-skill (this skill)Real Chrome/Edge via external bsk + extensionSites already signed into in the user's browser; setup/doctor: Extensions → BrowserSkill
computer-useNative macOS app UI (accessibility / screenshots)Controlling desktop apps — not web page automation

Prerequisites (user setup — do not dump shell unless asked)

If bsk --version fails or extension is disconnected, guide the user to:

OnMyAgent → Extensions → BrowserSkill

The panel has three actions: open Terminal to install CLI → open Chrome Web Store → recheck.

Do not paste long install docs into chat by default. Only mention the one-liner if they cannot use the desktop panel:

bash
curl -fsSL https://raw.githubusercontent.com/Tencent/BrowserSkill/main/install.sh | sh && bsk doctor

When to use

  • Sites the user is already signed into in Chrome/Edge
  • Navigate, snapshot, click, fill, screenshot against a real browser profile
  • Captcha / OTP / payment confirmation via bsk request-help

When NOT to use

  • Pure desktop app control → Computer Use
  • In-app webview / agent browser tools → browser-automation
  • Files, APIs, shell-only tasks with no browser
  • Credential harvesting (bsk evaluate on banking/SSO pages)

Mandatory lifecycle

1. bsk session start                 → capture 4-letter session id
2. every command: --session <id>
3. bsk session stop <id>             → REQUIRED (even on error)

Emergency: bsk session stop --all

Core loop

bsk navigate <url> --session <id>
bsk snapshot --session <id>          → aria tree with @e1, @e2 refs
bsk click @e3 --session <id>
bsk fill @e4 --value "…" --session <id>
bsk snapshot --session <id>          → re-snapshot after DOM change

Refs invalidate after navigation — always re-snapshot before the next interaction.

Observation priority

  1. bsk snapshot — default
  2. bsk get-html — only if snapshot is insufficient
  3. bsk screenshot — visual layout only when needed

Stop when the goal is met

Bounded goals only. When success is observed, bsk session stop immediately. On captcha/login blocks: bsk request-help, do not brute-force.

Red lines

  1. No token theft via evaluate on sensitive sites
  2. No long borrow of personal tabs across unrelated tasks
  3. Always session stop
  4. Prefer workspace artifacts over dumping secrets into chat

Docs

  • Upstream: https://github.com/Tencent/BrowserSkill
  • Any command: bsk <cmd> --help

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

See the full security report

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

See evaluation results

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/weaveq/onmyagent/browser-skill/badge)](https://verified-skill.com/skills/weaveq/onmyagent/browser-skill)

Something wrong with this skill or its page? Tell us.