Authenticated Web Research
Use this skill when the target site requires login, renders content dynamically, or is available only after the user signs in with their own account.
Hard Rule
Do not bypass access controls, paywalls, or anti-bot protections.
This skill is for user-authorized access only:
- - the user logs in with their own account
- the browser session stays local
- extraction continues only after access is legitimately available
When To Use
- - direct fetch returns login pages or partial shells
- search results show content exists, but direct fetch is blocked
- the site depends on client-side rendering
- the user explicitly wants help with a site they can access themselves
Workflow
1. Diagnose the failure mode
Classify the blocker:
- - login required
- JS-heavy rendering
- geo or locale mismatch
- thin snippet-only indexing
- temporary fetch incompatibility
2. Use browser-based loading
Before concluding the page is unavailable:
- - open the page in the local browser
- wait for client-side content to render
- inspect visible text, links, and network behavior where appropriate
- prefer the browser path over plain fetch for JS-heavy pages
3. Let the user complete login locally
If login is required and the user is authorized:
- - open the login flow in the local browser
- ask only for the minimum interaction needed, such as "please complete login in the opened page"
- do not ask for raw passwords or secrets in chat when browser login is possible
4. Continue within the authenticated session
After user login:
- - navigate to the target page
- search within the site or account area
- extract the needed facts, links, or structured results
- note which facts came from authenticated views
5. Prefer official post-login surfaces
If available, prefer:
- - account dashboards
- export pages
- official APIs
- RSS or feeds
- site search
- sitemaps
- downloadable reports
Output Pattern
Return:
- 1. what was reachable publicly
- what required authenticated access
- whether the user completed login locally
- what was extracted after authorized access
- what still remains unavailable
认证网络研究
当目标网站需要登录、动态渲染内容，或仅在用户使用自有账户登录后才可访问时，请使用此技能。
硬性规则
不得绕过访问控制、付费墙或反机器人保护。
此技能仅适用于用户授权访问：
- - 用户使用自有账户登录
- 浏览器会话保持本地运行
- 仅在合法获取访问权限后继续提取内容
使用时机
- - 直接抓取返回登录页面或部分空壳
- 搜索结果显示内容存在，但直接抓取被阻止
- 网站依赖客户端渲染
- 用户明确希望获取其可自行访问网站的相关帮助
工作流程
1. 诊断故障模式
对障碍进行分类：
- - 需要登录
- 重度JS渲染
- 地理位置或区域不匹配
- 仅提供简短摘要的索引
- 临时性抓取不兼容
2. 使用基于浏览器的加载
在判定页面不可用之前：
- - 在本地浏览器中打开页面
- 等待客户端内容渲染完成
- 在适当时检查可见文本、链接和网络行为
- 对于重度JS页面，优先使用浏览器路径而非直接抓取
3. 让用户在本地完成登录
如果需要登录且用户已获授权：
- - 在本地浏览器中打开登录流程
- 仅请求最低限度的交互操作，例如请在打开的页面中完成登录
- 在可通过浏览器登录时，不要在聊天中索要原始密码或密钥
4. 在认证会话中继续操作
用户登录后：
- - 导航至目标页面
- 在网站或账户区域内进行搜索
- 提取所需的事实、链接或结构化结果
- 记录哪些事实来自认证视图
5. 优先使用登录后的官方界面
如可用，优先使用：
- - 账户仪表盘
- 导出页面
- 官方API
- RSS或订阅源
- 站内搜索
- 站点地图
- 可下载报告
输出格式
返回：
- 1. 哪些内容可公开获取
- 哪些内容需要认证访问
- 用户是否在本地完成登录
- 授权访问后提取了哪些内容
- 哪些内容仍不可获取
