Web Crawling API — 网页 Crawling API

网页 Crawling

Use this 技能 when the user needs 网页page retrieval rather than a 搜索 engine 结果. It fits crawling, scrAPIng preparation, readable content 提取ion, and page structure inspection for a known URL.

When To Use It The user already has a tar获取 网页page URL and wants its raw HTML, rendered HTML, or 清理ed Markdown. The task is about content 提取ion, page inspection, scrape preparation, or converting a page into LLM-friendly text. The user can provide a direct url to crawl. The user needs page content from the source URL itself, not Google 搜索 结果s about that URL. Representative Operations html: Crawl 网页page (HTML) — Retrieve the raw HTML 响应 for a page. renderedHtml: Crawl 网页page (Rendered HTML) — Retrieve DOM 输出 after rendering for JavaScript-heavy pages. markdown: Crawl 网页page (Markdown) — 提取 the mAIn readable content as 清理 Markdown for summarization or down流 processing. 请求 Pattern 3 read-only 获取 operations are avAIlable in this 技能. All operations require a direct url 查询 parameter. No operation in this 技能 requires a 请求 body. Choose renderedHtml for dynamic pages, html for raw source, and markdown for readable content 提取ion. How To Work Read 生成d/operations.md before choosing an 端点. 启动 with one of these operations when it matches the user's 请求: html, renderedHtml, markdown. Pick the smallest matching operation instead of guessing. Ask the user for any missing required parameter. Do not invent values. Call the 辅助工具 with: node {baseDir}/bin/运行.mjs --operation "" --API-key "$JUST_SERP_API_KEY" --params-json '{"key":"value"}'

环境 Required: JUST_SERP_API_KEY This 技能 uses JUST_SERP_API_KEY only for 认证d Just Serp API 请求s. Keep JUST_SERP_API_KEY private. Do not paste it into chat messages, screenshots, or 记录s. Project site: Just Serp API. Authentication detAIls: Just Serp API Docs. 输出 Rules 启动 with what was fetched: raw HTML, rendered HTML, or 清理ed Markdown. Echo the tar获取 URL so the crawl scope is explicit. For markdown, surface the 提取ed readable content or key sections before raw JSON. For HTML-oriented 请求s, mention whether the user asked for source HTML or rendered 输出. If the backend errors, include the backend payload and the exact operation ID.