What Web search and extraction is for
Hermes documents separate tools for ranked search results and readable page extraction. Several backends are available, but support for search and extraction differs. This is a useful foundation for briefings and editorial research: search finds candidates, extraction helps inspect them, and the final output should preserve original links and dates. Use a browser only when interaction or rendering makes extraction insufficient.
What you need
- A chosen search backend and, when needed, a compatible extraction route.
- Provider authentication where required and a bounded research question with an explicit date window.
Set it up
- 1
Inspect the current web-tool backend through the documented configuration interface. Confirm both search and extraction capabilities rather than assuming one implies the other.
- 2
Test a known public page, then run a small query. Preserve URLs, publication dates where available, and retrieval gaps in the research record.
Check it works
- Open representative original sources and compare the extracted material with what the final summary claims.
- Separate publication date, event date, and collection time. Merge duplicate reports about the same underlying event.
Limits and failure points
- Search snippets may omit context; extracted text can miss tables, images, paywalled material, or dynamically loaded sections.
- Provider quotas and availability change. Failed retrieval should remain an explicit gap instead of being replaced by an unsupported summary.
Primary sources
Spotted something wrong or out of date? Tell us on X at @HermesWatcher (opens in a new tab) and we will fix it.