Pick sources and rank results for plain-language web search
Let Jev choose which sources and time window to search, then rank the results, returning links with scores instead of a written answer.
The use case
Researchers and agents running many web searches waste calls on the wrong sources and get long generated summaries they cannot verify. These tools take a plain-language query, have Jev choose the sources, time range and search terms, run the searches through an API, and then rank what returns. You get back ranked links and short excerpts, each with its score and adjustable filters, but no written answer, so a person reads the sources directly. It ships as a command-line tool and an MCP server so coding agents can call it.
Example workflow
- Accept a plain-language query from a user or agent
- Ask Jev Choice questions for sources and time window
- Run searches through a search API
- Ask Jev to score each result for relevance
- Return ranked links and snippets for a person to read
Tools
- Jev (TypeSafe AI)
- Search1API
- Claude Code
- MCP
Risks and guardrails
- Source lists limit what can be found
- Search API costs add up for agents
- Ranking cannot judge whether a source is accurate
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- Jev Search by superagents-lab on Made with Jev: madewithjev.com/builds/jev-search-web
- Jev Search: original post or repo by superagents-lab: github.com/superagents-lab/jev-search
- jev-search-mcp by MUKi on Made with Jev: madewithjev.com/builds/jev-search-mcp
- jev-search-mcp: original post or repo by MUKi: github.com/mukiwu/jev-search-mcp
- 中央社新聞 CNA TAIWAN NEWS (official page): ask.cna.com.tw/ask (checked 2026-11-19)