Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add aAAaqwq/AGI-Super-Team --skill browser-usegit clone --depth 1 https://github.com/aAAaqwq/AGI-Super-TeamWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/aaaaqwq/agi-super-team/browser-use)<a href="https://agentmods.dev/skills/aaaaqwq/agi-super-team/browser-use"><img src="https://agentmods.dev/badge/skills/aaaaqwq/agi-super-team/browser-use.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Tool Misuse · line 121 Tool defaults are unsafe or overly permissive (e.g. disabled TLS verification, no authentication, world-writable permissions). Unsafe defaults widen the attack surface.Fix: Override unsafe defaults with secure settings (verify=True, auth required, restrictive permissions). Review and harden all tool configurations.
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00057 | $0.02252 |
| Opus 5 | $0.00028 | $0.01126 |
| Sonnet 5 | $0.00011 | $0.00450 |
| Haiku 4.5 | $0.00006 | $0.00225 |
Grade A, and why
browser-use scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
subprocess.run(['pkill', '-f', 'chrome'], capture_output=True) How it starts
The opening of the file, as written. The whole thing — 365 lines — stays where its author put it; the contents beside it link to each section on GitHub.
browser-use 智能浏览器自动化
概述
browser-use 是一个 AI 驱动的浏览器自动化工具,它使用 LLM 来:
- 理解网页内容
- 智能决策下一步操作
- 自动完成任务
与 Playwright 的区别:
| 特性 | Playwright | browser-use |
|---|---|---|
| 控制方式 | 预编程脚本 | AI智能决策 |
| 适应性 | 页面变化需重写 | 自动适应 |
| Token消耗 | 较低但需调试 | 智能精简 |
| 复杂交互 | 需精确选择器 | 自然语言描述 |
| 维护成本 | 高 | 低 |
⚠️ 资源清理原则(强制)
所有涉及浏览器的 cron 任务完成后,必须自动关闭 Chrome 进程!
import asyncio
from browser_use import Agent
async def main():
agent = Agent(task="...", llm=llm)
result = await agent.run()
# ⚠️ 任务结束后必须显式关闭浏览器
if hasattr(agent, 'browser') and agent.browser:
await agent.browser.close()
# ⚠️ 推荐在脚本结束时强制清理残留进程
import subprocess
subprocess.run(['pkill', '-f', 'chrome'], capture_output=True)
return result
原因: 避免内存泄漏和资源占用,防止 Gateway CPU 100% 过载
安装状态
✅ 已安装:
- browser-use 0.11.11
- browser-use-sdk 2.0.15
快速开始
基本用法
import asyncio
from browser_use import Agent
from langchain_openai import ChatOpenAI
async def main():
agent = Agent(
task="打开 polymarket.com,查看 Fed 利率市场",
llm=ChatOpenAI(model="gpt-4o"),
)
result = await agent.run()
print(result)
asyncio.run(main())
使用自定义 LLM(推荐配置)
⚠️ 重要:your-provider API 是 Anthropic 格式,不是 OpenAI 格式!必须使用 ChatAnthropic。
from browser_use.llm.anthropic.chat import ChatAnthropic
# your-provider API(Anthropic 兼容)✅ 推荐
llm = ChatAnthropic(
model="claude-sonnet-4-6",
base_url="https://your-anthropic-proxy.example.com", # 注意:不加 /v1
api_key="your-api-key", # 或从 pass show api/your-provider 获取
)
agent = Agent(
task="你的任务",
llm=llm,
)
# ❌ 错误用法:不要用 ChatOpenAI + your-provider
# from browser_use.llm.openai.chat import ChatOpenAI # 这个不行!your-provider 不支持 OpenAI 格式
# 如果使用 OpenAI 兼容 API(如 Provider-B),用 ChatOpenAI:
from browser_use.llm.openai.chat import ChatOpenAI
llm = ChatOpenAI(
model="gpt-4o",
base_url="https://ai.9w7.cn/v1",
api_key="your-api-key",
)
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 365 lines · 57 tokens per session scan A c6f2e1e05672
browser-use is a skill published in the GitHub repository aAAaqwq/AGI-Super-Team (91 stars, last pushed yesterday), licensed MIT. It adds 57 tokens to every session and 2,252 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.
Other skills, from other repositories
browser-check
Drive a real browser and come back with a MEASUREMENT rather than an impression - console errors and >=400 responses as counts, computed styles as JSON when appearance is disputed. Use when work touches UI, when a review must verify one, or when fidelity-gate asks for its measurement.
qa
QA testing skill with real browser automation. Use when asked to "test this site", "QA this page", "check for visual bugs", "verify the deploy", or when Hydra needs browser validation for UI changes. Requires the browse binary.
chrome-devtools
Drive the machine Chrome debug browser via OpenClaw-managed MCP (chrome-devtools). Use for page navigation, snapshots, screenshots, clicks, forms, console/network inspection — not for host shell risk.
mission-control
Interact with Mission Control — AI agent orchestration dashboard. Use when registering agents, managing tasks, syncing skills, or querying agent/task status via MC APIs.
browser
Browser automation via the agent-browser CLI. Use when the user needs to drive websites or Electron desktop apps — navigating, filling forms, clicking, screenshots, extracting data, testing web apps, visual UI checks, the Pi Dashboard's Electron shell, or the user's own logged-in browser (SSO/2FA sites). Triggers…
prompt-tuning
Tune a prompt, or anything whose quality is measured by non-deterministic model output, without chasing noise - a noise baseline before the first edit, medians over repeated runs, enforcement AFTER generation rather than in the wording. Use when iterating on prompts or model-judged output.