dsh-cross-search

A single search engine can return results, but it is hard to tell whether those results are independently verified or merely an isolated entry in its index. dsh-cross-search turns this judgment into a measurable process: it sends the same query to multiple independent engines in parallel and uses the number of agreeing engines as the sole confidence signal.

Under the hood, the plugin fetches from Brave, Tavily, DuckDuckGo and Wikipedia (with optional Bing and Mojeek) concurrently, deduplicates returned URLs, strips tracking parameters, and classifies every result into one of three tiers. HIGH means more than half of the running engines agree, MEDIUM means at least two agree, and LOW means a single engine found the match. Each result carries its engine provenance, and the final report states exactly which engines participated and which were skipped due to missing API keys, keeping the process fully transparent.

It suits anyone who needs cross-verified information: researchers who do not want a single index to anchor their conclusions, writers producing fact-checked or technical documents, and developers who want their AI search pipeline to carry a built-in confidence filter. Setup is straightforward: supply the API keys you have, and DuckDuckGo and Wikipedia join the pool without any key, so the plugin works out of the box.