JevGrep (Arifette)
CLI and MCP server where Jev scores code fragments for a behaviour question and returns exact excerpts with line numbers.
Overview
JevGrep (Arifette) helps coding agents find code by what it does when they do not know the file name or symbol. Ask a question such as where is session expiry handled, and it scans the authorised repository, asks Jev to score all eligible fragments, and returns the original source excerpts with paths and line numbers so the agent can read those files in detail with less exploratory context. It is meant to complement exact tools such as ripgrep, and searches any valid UTF-8 text file regardless of language.
It emphasises consent and inspection before sending code: init asks whether eligible excerpts may be sent to the chosen provider and leaves remote evaluation disabled unless you agree, creates a .jevgrepignore alongside respecting .gitignore, and offline doctor and inspect commands show provider and credential state, the authorised root, limits, cache, which fragments are eligible and how much work a search would do. Providers include TypeSafe, Vercel AI Gateway and OpenRouter. It is available as a CLI and a local MCP server, requires Node.js 24, and is published under a scoped npm name since the plain name belongs to another project. MIT licensed.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How relevant is this code fragment to the behaviour question? | — |
How it works
- Init a profile and consent to remote evaluation
- Inspect eligible fragments offline
- Search: Jev scores fragments
- Return excerpts with paths and line numbers to the agent
Strengths
- Explicit consent before sending code
- Offline inspect and doctor commands
- Works across languages
- CLI and MCP server
Caveats
- Requires Node.js 24
- Scanning many fragments has a cost
Getting started
npm install -g @nassim-arifette/jevgrep, then jevgrep init and `jevgrep search --query "Where is session expiry handled?"`.At a glance
- Type
- MCP server
- Status
- beta
- Language
- TypeScript
- Runtime
- Node CLI / MCP server
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Jev-powered semantic code search for coding agents — find behavior across repositories via CLI or MCP, with exact source excerpts and line numbers.
logicrw/awesome-jev-projectsSemantic code search for agents, as a CLI or MCP server: ask what the code does and get source excerpts with paths and line numbers.
MrJev/awesome-jev
Frequently asked questions
Will it send my code without asking?
No, remote evaluation stays disabled until you consent during init or edit the profile.
Which npm package should I install?
@nassim-arifette/jevgrep; the unscoped name belongs to a different project.
When is plain rg better?
When you already know the symbol or literal you need.
Related projects
Deterministic code-graph context for coding agents on Postgres, AGE and pgvector, with optional Jev or local re-selection of symbols.
MCP serverbetaPythonSingle 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 serverbetaPythonCLI and MCP server giving coding agents Jev-powered semantic search, focused reads and list picking in about a second for about a cent.
BenchmarkbetaPythonMCP server and semantic index that tags shadcn-format components once and matches a plain-language brief to them deterministically.
MCP serverbetaTypeScriptMCP server over Apple developer docs and WWDC transcripts that uses hybrid search, then Jev relevance scores to rank results.
MCP serverproductionTypeScriptLocal long-term memory for coding agents with an optional Jev reranker that orders recalled memories.
ProjectproductionTypeScript