qsearch
For agents that keep hallucinating because they're reading snippets instead of full pages, this plugs in multi-engine search with trust scoring built from repeated sweeps. You get full content extraction, per-URL provenance showing which engines agreed (Google, Brave, DuckDuckGo, Qwant via SearXNG), and a local Meilisearch corpus that ranks sources by how often they survive independent queries. The sweep endpoint batches queries with priority routing (free SearXNG for broad, paid Brave for focused, academic APIs for papers), the corpus viewer shows trust scores across topics, and everything runs self-hosted with your own API keys. Built for research sprints where you need to see what multiple engines confirm over time, not just what ranks first today.
- GitHub stars
- 1
- tools
- 9
- transport
- HTTP
- last updated
- Apr 29, 2026
Tools
9 toolsPublic tool metadata for what this MCP can expose to an agent.
web_searchweb search via Brave or SearXNG
web search via Brave or SearXNG
No parameter schema in public metadata yet.
sweepbatch research sweep with multi-engine attribution
batch research sweep with multi-engine attribution
No parameter schema in public metadata yet.
academic_searchpeer-reviewed papers via arxiv + PubMed + Semantic Scholar
peer-reviewed papers via arxiv + PubMed + Semantic Scholar
No parameter schema in public metadata yet.
sweep_contextPhase 3 local LLM page extraction (free, Ollama)
Phase 3 local LLM page extraction (free, Ollama)
No parameter schema in public metadata yet.
verify_citationdoes the cited URL actually support the claim? (Supported/Partial/Unsupported/Fabricated)
does the cited URL actually support the claim? (Supported/Partial/Unsupported/Fabricated)
No parameter schema in public metadata yet.
economy_reportcost breakdown vs all-Brave baseline
cost breakdown vs all-Brave baseline
No parameter schema in public metadata yet.
index_researchindex local .md files by glob
index local .md files by glob
No parameter schema in public metadata yet.
news_searchnews search (Brave key required)
news search (Brave key required)
No parameter schema in public metadata yet.
context_searchdeep page content (Brave key required)
deep page content (Brave key required)
No parameter schema in public metadata yet.
Sponsors
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any languageReadme
qsearch
I built this for my own daily research. After running 100+ research sprints, my agent kept hallucinating because it read 200-char snippets. qsearch gives it full content with multi-engine provenance — running locally, owned by me.
AI agents lose 17–33% of facts to hallucination because they read 200-character snippets, not full pages (Stanford 2024). Existing search APIs hide which engines agreed on a result. Existing knowledge graphs are enterprise-priced or vendor-locked.
Details
- Repository
- github.com/theyahia/qsearch
- Website
- qsearch.pro
- Category
- Search & Web Crawling
- Transport
- HTTP
- Registry
- MCP registry, active
More from theyahia
Related Search & Web Crawling MCP servers
View all
Reftrixmcptkmd/reftrixmcpWeb design analysis with 26 MCP tools: layout, motion, quality, semantic search via pgvector1stars
Aamangeldi Dad Jokes Mcpaamangeldi/dad-jokes-mcpGet a random dad joke or search by keyword to fit any moment. Retrieve specific jokes by ID for re…0stars
Cindyloo Dropbox Mcp Servercindyloo/dropbox-mcp-serverSearch, browse, and read your Dropbox files. Find documents by name or content, list folders, and…0stars
xurprisenimo1987/xurprise-mcp-docsAgent-native commerce for SEA/China/global. 43 brands. 6 tools. 9-language search.0stars
Upnote Lens Mcpawesomehye/upnote-lens-mcpRead, search, and summarize your UpNote notes — and create new ones.0stars