jevgrep
Jev-powered semantic code search for coding agents — find behavior across repositories via CLI or MCP, with exact source excerpts and line numbers.
Added to Jevfast
How it uses Jev
JevGrep scans eligible repository fragments and asks Jev to score their relevance to a natural-language code question. It returns matching original excerpts with paths and line numbers for the calling agent to inspect.
What Jev decides
Code excerpt relevance requestExample answers · not a recorded Jev response · Source ↗
Question 1 · f-one
YOUR APP
INSTRUCTION
Does this source excerpt contain concrete evidence useful to the search question?
STATE
The shared state contains the search question and criterion. A question for each candidate excerpt uses that item’s runtime ID; f-one is an ID used in the repository test fixture.
JEV · NOUL
YesNo
App workflow
- Describe the code sought
Provide a natural-language question such as where session expiry is handled.
- Score repository fragments
Search the authorized repository and evaluate eligible fragments for relevance.
- Inspect the matches
Return source excerpts, paths, and line numbers so the agent can read the surrounding code.
Why it is interesting
The search works from what code does rather than requiring a known symbol or filename, and provides both a CLI and a local MCP interface for coding agents.