| name | web-research |
| description | Web search, data extraction, and structured report generation from online sources. |
Web Research
Research Workflow
- Define scope — Identify what information is needed and from what sources
- Search & fetch — Use web_fetch to gather data from relevant URLs
- Extract & validate — Pull specific data points, cross-reference sources
- Synthesize — Combine findings into a structured report
- Save — Write results to a well-formatted file
Web Fetching Strategies
API-first approach
Many sites offer structured JSON at predictable URLs:
https://wttr.in/City?format=j1 — weather data
https://api.example.com/v1/data — REST APIs
- Add
?format=json or accept headers for JSON
Fallback: HTML parsing
When no API exists, fetch the HTML and extract data using patterns in the response text.
Structured Reports
When creating research reports, use this structure:
# [Topic] Research Report
## Executive Summary
[2-3 sentence overview of key findings]
## Findings
### [Category 1]
- Key point with supporting detail
- Data point with source
### [Category 2]
...
## Comparison Table
| Criterion | Option A | Option B | Option C |
|-----------|----------|----------|----------|
| Feature 1 | ... | ... | ... |
## Market Trends / Analysis
[Broader context and patterns]
## Conclusion
[Actionable takeaways]
Data Accuracy
- Prefer primary sources over aggregators
- Note when information may be outdated
- Use multiple sources for critical facts
- Clearly attribute sources when available
- If web access is unavailable, use knowledge but note the limitation