Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

jegrep

Rust semantic grep that scores live repository files and line ranges with Jev yes/no probabilities, with no index or embeddings.

ProjectbetaMIT
View on GitHub ↗

Overview

jegrep is a natural-language search tool that works like grep. You describe what you want, such as where do we verify JWT tokens, and it searches the live tree on every run with no embeddings, index or daemon, returning files and line ranges as dirname/filename:first-last references with merged adjacent passages. Each path gets an absolute yes/no probability, so thresholds are meaningful and batches are comparable.

It is aimed at developers and coding agents, with --json for scripts, --compact for a token-lean digest, and a benchmark harness for regressions. The README cites Jev pricing of $0.042 per million input tokens and estimates a typical search over a few thousand files at about $0.005. It can use OpenRouter or TypeSafe directly, prefers OpenRouter when both keys exist, fails over on auth, rate-limit, timeout and server errors, and offers --only to refuse failover. Binaries are provided for Linux, macOS and Windows, and it is MIT licensed.

What the decision model decides

TypeQuestionAnswers
Yes / noDoes this file or passage match the described concept?(inferred)yes · no

How it works

  1. Run jegrep with a natural-language query
  2. Walk the live tree and chunk files
  3. Ask Jev a yes/no per chunk via OpenRouter or TypeSafe
  4. Return matching files and line ranges above threshold

Strengths

  • No index to build or keep fresh
  • Calibrated probabilities make thresholds meaningful
  • README estimates about $0.005 per typical search
  • Automatic provider failover

Caveats

  • Sends code to a hosted API
  • Linux binaries need glibc 2.39+

Getting started

`cargo install jegrep`, then set OPENROUTER_API_KEY or TYPESAFE_API_KEY.

At a glance

Type
Project
Status
beta
Language
Rust
Runtime
CLI
Has code
yes
Has tests
yes
License
MIT

Use cases and integrations

How curated lists describe it

  • Rust semantic grep that scores live repository files and ranges with Jev probabilities, without an embedding index or background daemon.

    cobanov/awesome-jev

  • Semantic grep with no embeddings, index, or daemon: it searches the live tree on every run and matches concepts rather than strings.

    MrJev/awesome-jev

Frequently asked questions

What does jegrep decide?

Whether each file or passage matches your natural-language description, with a probability.

Does it need an index?

No. It searches the live tree on every run.

What does a search cost?

The README estimates about $0.005 for a few thousand files.

Related projects

  • askgrep

    Semantic grep that asks Jev a yes/no about every function for questions no regex can express, with cost reporting.

    Projectbeta
  • BGTS Context Engine

    Deterministic code-graph context for coding agents on Postgres, AGE and pgvector, with optional Jev or local re-selection of symbols.

    MCP serverbetaPython
  • claude-jev

    Claude Code plugin that adds Jev as an independent judge for review findings, debugging hypotheses, design options and search hits.

    Pluginbeta
  • codex-jev

    Retrieval layer for Codex that finds evidence locally, optionally ranks sanitized excerpts with Jev, and returns bounded, cited evidence.

    Projectbeta
  • dsh-jev-grep

    DeepSeek Harness jev_search tool that runs ripgrep, scores every hit line with Jev, and returns only the top rows above a cut.

    PluginbetaJavaScript
  • Every

    CLI that asks one yes/no question of every function in a codebase with Jev and ranks them by probability: grep whose pattern is a question.

    CLIbetaPython