← 返回
未分类 中文

Browser Task Runner

Plan and execute repeatable browser-based workflows such as opening sites, logging in, navigating dashboards, filling forms, exporting data, taking screensho...
规划并执行可重复的浏览器工作流,如打开网站、登录、浏览仪表盘、填写表单、导出数据、截图。
benboby benboby 来源
未分类 clawhub v1.0.0 1 版本 100000 Key: 无需
★ 0
Stars
📥 96
下载
💾 1
安装
1
版本
#latest

概述

Browser Task Runner

Treat browser work as fragile. Prefer deterministic, repeatable flows over improvising clicks.

Workflow

  1. Confirm the target site and exact outcome.
  2. Identify the available execution path:
    • direct web fetch/search for read-only public pages
    • local browser automation tooling if available
    • AppleScript/UI scripting as a fallback for simple fixed flows
  3. Check whether the task needs authentication, MFA, CAPTCHA, or sensitive confirmation.
  4. For destructive or account-changing actions, ask before proceeding.
  5. Execute the smallest reliable flow and report where manual takeover may still be needed.

Good fits

  • open a fixed set of URLs
  • log into a routine internal tool
  • navigate to a report page
  • export a CSV or PDF
  • fill a repetitive form
  • capture a screenshot of a page state
  • scrape visible page data into a summary

Fragile cases

Be cautious when the task involves:

  • CAPTCHA
  • MFA prompts
  • anti-bot protections
  • drag-and-drop heavy UIs
  • unpredictable popups
  • account settings or payment actions

In those cases, prefer a hybrid flow: automate the boring parts and let the user finish the sensitive step.

Execution policy

Prefer this order:

  1. Read-only web tools for public pages
  2. Existing local automation tools/frameworks
  3. macOS UI scripting for narrow, stable workflows

Do not pretend browser control exists if the tooling is absent. State what is missing and what can still be automated.

Output pattern

Report using:

  • Goal
  • Automation path
  • What succeeded
  • Where it may break
  • Next improvement

Keep website-specific details in scripts or references if this skill grows beyond a generic runner.

版本历史

共 1 个版本

  • v1.0.0 当前
    2026-06-07 06:52

安全检测

腾讯云安全 (Keen)

队列中

腾讯云安全 (Sanbu)

队列中

🔗 相关推荐

life-service

Wechat Summary

benboby
在保护隐私的前提下,总结微信对话、筛选未读消息、识别需要回复的内容,并提供简洁的回复建议。
★ 0 📥 174
ai-agent

Self-Improving + Proactive Agent

ivangdavila
自我反思+自我批评+自我学习+自组织记忆。智能体评估自身工作、发现错误并持续改进。
★ 1,412 📥 325,384
ai-agent

self-improving agent

pskoett
记录自身发现以实现自我改进的技能
★ 4,124 📥 863,514