JevFind
Rust semantic code search: Jev scores file paths, then code windows in the survivors, and prints matching line ranges.
Overview
Jev Code Finder (JevFind) is a semantic code search tool. You describe a concept in plain English, such as where does user JWT authentication happen, and it returns relevant files, line ranges, confidence scores and source snippets. Unlike text search, it finds meaning, so a query like where are permissions enforced can surface middleware, guards, policy checks and configuration without knowing their names.
It walks the repository respecting .gitignore and excluding hidden, dependency, cache and build directories, asks Jev to score each source path for likely relevance, opens only paths above a file threshold, splits them into overlapping, size-bounded windows, sends windows in bounded batches and prints matches above a match threshold. The path filter keeps unnecessary code out of later requests; setting the file threshold to zero favours recall over speed. Transient errors are retried up to three times with exponential backoff. It installs via Homebrew or cargo, ships an agent skill, needs Rust 1.88+ and a TypeSafe key, and is MIT licensed.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How likely is this file path to be relevant to the query? | — |
| Score | How relevant is this code window to the query? | — |
How it works
- Walk the repo respecting ignore rules
- Jev scores file paths; keep those above --file-threshold
- Split kept files into code windows and score in batches
- Print windows above --threshold with snippets
Strengths
- Finds code by meaning, not names
- Two-stage filtering keeps cost down
- Retries with backoff
- Agent skill for coding agents
Caveats
- Sends code snippets to TypeSafe
- File-level filter may miss relevant files unless threshold lowered
Getting started
brew tap Peu77/tap && brew install --HEAD jev-code-finder, or `cargo install --path .`, then set TYPESAFE_API_KEY.At a glance
- Type
- Project
- Status
- beta
- Language
- Rust
- Runtime
- CLI / agent skill
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Fast semantic code search powered by Jev. Find the relevant files, line ranges, and snippets
logicrw/awesome-jev-projects
Frequently asked questions
How is it different from grep?
It searches for meaning, so you need not know function or file names.
How does it keep costs down?
It scores file paths first and only opens files above a threshold.
How do I maximise recall?
Set --file-threshold 0.
Related projects
Deterministic code-graph context for coding agents on Postgres, AGE and pgvector, with optional Jev or local re-selection of symbols.
MCP serverbetaPythonClaude Code plugin that adds Jev as an independent judge for review findings, debugging hypotheses, design options and search hits.
PluginbetaRetrieval layer for Codex that finds evidence locally, optionally ranks sanitized excerpts with Jev, and returns bounded, cited evidence.
ProjectbetaPipe grep or file lists into Jev to get every candidate scored, sorted and cut at a calibrated threshold, so agents read only what matters.
ProjectbetaTypeScriptsys1grep: a zero-dependency grep that finds lines by meaning, scoring each line with Jev, with AND/OR/NOT and cross-language queries.
ServicebetaJavaScriptSingle MCP tool that indexes a repo and has Jev rank a shortlist so a coding agent reads the right files and lines instead of grepping.
MCP serverbetaPython