jev-search
Search the web with TypeSafe's Jev: source selection, query understanding and relevance ranking.
Overview
jev-search (github.com/superagents-lab/jev-search) is an MIT TypeScript worker that accepts plain-language questions, lets Jev choose providers, time ranges, and keyword variants, calls Search1API for raw results, then Jev Scores relevance before returning links and snippets only. Try the hosted UI at jev.s1.dev or deploy the worker with your keys.
Problem: Agent search demos either hallucinate answers or ship a single hard-coded Google query. Builders want ranked evidence with explicit source planning.
Built for: Engineers self-hosting Cloudflare Workers search who want Jev to plan queries and score hits without a chat summarizer on top.
First indexed on Jev Directory: 2026-09-18
Creator and team
- Name
- Superagents Lab
- Organization
- Superagents Lab
- Handle
- @superagents-lab
How Jev is used
- Role in the product flow
- Query planning and per-hit relevance ranking over Search1API results
- Primitives
- ChoiceScoreNoul
- State in
- User question text, configured provider list, and candidate search parameters (sources, recency windows, query variants).
- Decision out
- Planned fetch plan (which sources and terms to run) plus per-result relevance scores that gate what surfaces in the UI.
- User submits natural language query at jev.s1.dev or the worker API
- Jev Choice selects sources, time filters, and query reformulations
- Search1API returns raw hits without generated prose
- Jev Score ranks snippets; UI shows links and excerpts only
jev-search treats Jev as a planner and reranker, not an answer bot. The worker keeps HTTP boundaries crisp: one planning pass decides how to spend Search1API quota, then structured scoring trims noise before anything renders. That matches the directory pattern of typed gates instead of free-form chat. There is no public X launch clip tied to this repo in this batch, so the thick page leans on README architecture and the live demo. If you need channel-scoped search with caption jumps, compare the My First Million demo listing; jev-search is general web retrieval.
Sourced performance claims
- Public GitHub repo superagents-lab/jev-search had about four hundred twenty seven stars when this listing was drafted.Source: GitHub star count September 2026
Features and stack
Features
- MIT Cloudflare Workers deployment path
- Search1API integration with BYOK configuration
- No generated answer layer in the default UX
- Hosted demo at jev.s1.dev
Stack
- TypeScript
- Cloudflare Workers
- Search1API
- TypeSafe Jev
Pricing: Open source worker; Search1API and TypeSafe usage bill to your keys on self-hosted runs.
Links
FAQ
- Does jev-search write summaries?
- The product positioning is links plus snippets ranked by Jev. It is not a chat completion wrapper over search results.
- Can I self-host?
- Yes. README documents Cloudflare Workers deployment with your Search1API and TypeSafe credentials.
- How is this different from the My First Million demo?
- mfm-jev-search scopes retrieval to one YouTube catalog with hybrid recall. jev-search targets general web search planning via Search1API.
Related learn guides
Original Jev guidance that pairs with this product pattern.
- Jev use cases
The patterns builders actually search for: moderation, routing, triage, RAG verify, and agent gates.
- Jev vs LLM classification
When to gate with System One probabilities instead of asking a chat model to label things.
Related products
Hand-picked neighbors with rich profiles or overlapping tags.