| name | web-research |
| description | 网络研究 — 使用搜索引擎和网页抓取获取外部信息，支持多源交叉验证 |
| version | 1.0.0 |
| license | MIT |
| metadata | {"author":"blackplume233","actant-tags":"research,web,search,researcher","layer":"auxiliary","agent":"researcher"} |
Web Research
Overview
Use search engines and web scraping to obtain external information; support multi-source cross-validation.
Search Strategy
- Query formulation: Expand user query with synonyms, related terms; use Boolean operators when supported
- Source selection: Choose engines (Google, Bing, DuckDuckGo) and verticals (docs, GitHub, Stack Overflow) by topic
Multi-Engine Search
- Execute parallel searches across engines; merge and deduplicate results
- Handle rate limits and fallbacks; respect robots.txt
Content Extraction and Summarization
- Extract main content from pages; strip ads, nav, footers
- Summarize long articles; preserve key facts and citations
- Support structured data (JSON-LD, meta tags) when available
Source Credibility Assessment
- Score sources by domain authority, recency, citation count
- Flag low-quality or suspicious sources for user awareness
Result Deduplication
- Detect near-duplicate content across sources; cluster by similarity
- Present unique insights; cite multiple sources for corroboration
