web-research

Source-attributed research summaries are generated by searching web content, extracting relevant information, cross-referencing claims across multiple sources, and outputting findings in a consistent template with conflict flagging.

Fact checking and synthesizing online information is best handled through a structured search strategy that breaks questions into sub-queries, searches each independently, records URLs and publication dates, and cross-references claims across multiple sources to flag conflicts.

Multi-step web investigations with conflict flagging require breaking research questions into sub-queries, extracting content with freshness checks, and synthesizing findings while explicitly flagging conflicting information across sources in a consistent output template.

Automated web research supports overnight agent research workflows by converting open-ended questions into reliable, source-attributed summaries through structured search, content extraction with publication dates, and synthesized conclusions with explicit source attribution.

Source attribution and recency checking work for policy reviews and product decisions by recording publication dates during content extraction, cross-referencing claims across multiple sources, and producing evidence-backed overviews with explicit per-finding relevance.

Limitations of web research include dependency on search and fetch capabilities to retrieve web content, requiring multiple independent sources to cross-reference claims, and the need to manually review flagged conflicts within the synthesized output template.