page-agent-browser通过 page-agent CLI(CDP)驱动浏览器;CLI 当前仅从 GitHub Release 的 .tgz 安装(npm 官方包尚未发布)。
Install via ClawdBot CLI:
clawdbot install sdyuyouth/page-agent-browserGrade Fair — based on market validation, documentation quality, package completeness, maintenance status, and authenticity signals.
Potentially destructive shell commands in tool definitions
Exec(Calls external URL not in known-safe list
https://github.com/sdyuyouth/page-agent-cli/releasesAI Analysis
The skill primarily orchestrates a local browser automation CLI via Bash and does not exfiltrate user data to external servers. The only external reference is to a GitHub releases page for downloading the CLI tool, which is consistent with the stated purpose. No hidden instructions, credential harvesting, or obfuscation patterns are present.
Audited May 10, 2026 · audit v1.0
Generated Oct 2, 2026
QA 团队用 page-agent 通过 CDP 连接本机 Chrome/Edge,按 state 返回的索引执行 click、input、hover 等原语回放用户流程。每次发布前跑同一套脚本,对比 eval 返回值确认关键元素和接口未被破坏。
运营人员把 Excel/CRM 数据整理成 JSON,用 input、select、upload 批量填写政务、供应链或保险类 Web 表单。遇到复杂上传区时先用 eval/click 打开,再用指针索引让 CLI 自动匹配最近的 input type=file。
自动化工程师用 teach 录制人工操作流程,确认写入 Agent 经验后从 stdout 拿到完整 JSON,落到 platforms/<site>/recipes/*.md。之后再遇到同一站点的新任务,Agent 先读该站点知识库再执行,减少反复探索。
财务或风控人员用 tabs list + goto 遍历多个内网系统页面,用 eval 提取 DOM 中的订单号、金额和状态字段。脚本把结果合并成 CSV,按规则比对差异并生成对账报告。
产品团队用 page-agent 的 teach 阻塞式浮窗跟随真人操作,自动把每步 click、input、hover 和 state_refresh 记录成结构化步骤。录制结果用于生成新员工上手指南和产品功能演示脚本。
咨询或外包公司给金融机构、制造业客户搭建基于 page-agent 的网页自动化流程,按项目交付脚本、站点经验库和运行监控。客户无需自建 Agent 团队即可获得稳定的浏览器自动化能力。
把 page-agent 的 CDP 控制能力封装成多租户 SaaS,提供可视化流程编辑、站点元素库和定时任务调度。客户在自己受控的浏览器环境里运行流程,平台只负责编排、日志和异常告警。
围绕 teach 产出的站点 recipes、elements.md 和 checkpoint 恢复模板,建立可按域名和任务检索的技能市场。开发者出售自己沉淀的高质量流程经验,买家直接导入自建 platforms 目录复用。
💬 Integration Tip
先确认本机 Chrome/Edge 已用 --remote-debugging-port 启动且未占用用户数据目录,再用 curl http://localhost:9222/json/version 验证 CDP 连通;teach 会长时间阻塞,务必单独设置宽松超时并在 exit 0 后读取 stdout JSON。
Scored Oct 2, 2026
A fast Rust-based headless browser automation CLI with Node.js fallback that enables AI agents to navigate, click, type, and snapshot pages via structured commands.
Uses a headless browser to navigate web pages, interact with elements, and extract clean, readable text content from URLs.
Headless browser automation CLI optimized for AI agents with accessibility tree snapshots and ref-based element selection
High-performance browser automation for heavy scraping, multi-tab management, and precise DOM extraction. Use this when you need speed, reliability, or advanced state management (cookies/local storage) beyond standard web fetching.
Ultimate stealth browser automation with anti-detection, Cloudflare bypass, CAPTCHA solving, persistent sessions, and silent operation. Use for any web automation requiring bot detection evasion, login persistence, headless browsing, or bypassing security measures. Triggers on "bypass cloudflare", "solve captcha", "stealth browse", "silent automation", "persistent login", "anti-detection", or any task needing undetectable browser automation. When user asks to "login to X website", automatically use headed mode for login, then save session for future headless reuse.
Browser automation CLI (browser-act) for AI agents. MUST trigger when: (1) user mentions 'browser-act' in any form, or user needs to: (2) open/visit/browse/c...