Feature Request: Add `renderMode` to `fetchWebContent` for explicit Playwright/browser rendering | 新功能：为 fetchWebContent 增加 renderMode，支持显式使用 Playwright / 浏览器渲染

[Aas-ee]

Feature Request: Add renderMode to fetchWebContent for explicit Playwright/browser rendering

Background

Currently, fetchWebContent mainly uses the request-based path and only falls back to browser-assisted fetching in limited cases, such as bot challenge pages, very low
extracted content, or some 401/403/429 responses.

However, some websites rely heavily on client-side JavaScript rendering. For these pages, request-based fetching may return an empty shell or incomplete content, and the current
automatic fallback may not always trigger.

Proposal

Add an explicit renderMode parameter to fetchWebContent:

renderMode?: "request" | "auto" | "browser"

Suggested behavior:

• request: use the current request-based behavior only
• auto: keep the current behavior, request first with browser fallback when needed
• browser: directly use Playwright/browser rendering to load the page and extract content

The default should remain auto or the current behavior, so existing users are not affected.

Expected Benefits

• Allows users and agents to explicitly fetch JS-rendered pages with Playwright
• Avoids overloading searchMode, which currently only affects Bing search
• Keeps normal request-based fetching fast and lightweight by default
• Makes browser rendering an intentional opt-in instead of an implicit fallback

Safety Notes

The implementation should still preserve the existing safety checks:

• Validate the initial URL before navigation
• Validate the final URL after browser navigation
• Keep blocking private/local network targets by default
• Return a clear error when Playwright or a browser executable is unavailable

---

功能请求：为 fetchWebContent 增加 renderMode，支持显式使用 Playwright / 浏览器渲染

背景

目前 fetchWebContent 主要走 request 抓取路径，只会在少数场景下尝试浏览器兜底，例如遇到 bot challenge、提取内容过少，或者部分 401/403/429 响应。

但有些网站严重依赖前端 JavaScript 渲染。对于这类页面，request 抓取可能只能拿到空壳 HTML 或不完整内容，而当前自动兜底不一定总能触发。

建议方案

给 fetchWebContent 增加一个显式参数：

  renderMode?: "request" | "auto" | "browser"

建议行为：

• request：只使用当前 request 抓取逻辑
• auto：保持当前行为，先 request，必要时浏览器兜底
• browser：直接使用 Playwright / 浏览器加载页面并提取内容

默认值应保持为 auto 或当前行为，避免影响现有用户。

预期收益

• 允许用户和 Agent 显式抓取 JS 渲染页面
• 避免复用 searchMode，因为它目前只影响 Bing 搜索
• 普通网页默认仍然保持轻量、快速的 request 抓取
• 浏览器渲染变成明确的用户选择，而不是隐式兜底

安全注意事项

实现时仍然应保留现有安全检查：

• 浏览器导航前校验初始 URL
• 浏览器导航后校验最终 URL
• 默认继续阻止本地 / 内网地址
• Playwright 或浏览器不可用时返回清晰错误