dumplingai-web-research

当用户希望通过DumplingAI进行网络研究、搜索结果获取、引用来源查找或快速主题发现，而不是手动浏览或直接搜索API集成时。此外，当用户提到谷歌搜索、SERP结果、网络搜索、寻找来源、主题研究、竞争对手研究、答案来源、Serper、Perplexity，或者想要比较网络上关于某事的说法时也使用此功能。每当任务首先是搜索，然后是检查或综合结果时，请使用此功能。对于在发现后抓取特定页面或网站，请参阅dumplingai-scraping-extraction。

DumplingAI Web Research

Allowed Commands

dumplingai catalog search <prompt>
dumplingai catalog details <type> <id>
dumplingai run <type> <id> --input '<json>'

Workflow

Translate the task into a short keyword query such as google search, web search, or perplexity search.
Run dumplingai catalog search "<job>".
Inspect the best candidate with dumplingai catalog details <type> <id>.
Execute the search capability or endpoint with a focused query.
Save large outputs under .dumplingai/ and read them incrementally.
If the user needs page-level content extraction after discovery, hand off to dumplingai-scraping-extraction.

Search Phrases

google search
web search
serper search
perplexity search
news search

Output Strategy

dumplingai catalog search "google search" > .dumplingai/catalog-search.json
dumplingai catalog details capability google_search > .dumplingai/google-search.json
dumplingai run capability google_search --input '{"query":"best AI meeting notes tools for startups"}' > .dumplingai/search-results.json
head -60 .dumplingai/search-results.json
rg '"title"|"url"|"snippet"|"results"' .dumplingai/search-results.json