Skip to content
Back to search

jevgrep

Find code by asking what it does. A CLI for coding agents that uses Jev to discover relevant files and source context.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/dzhng/jevgrep with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/dzhng/jevgrep'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/dzhng--jevgrep/badge.svg)](/listings/dzhng--jevgrep)

HTML

<a href="/listings/dzhng--jevgrep"><img src="/api/listings/dzhng--jevgrep/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Installing the CLI alone does not teach your coding agent to use it. **Install the skill as well**, from the project where your agent works:

jg skill

The installer detects your coding agents (Claude Code, Codex, OpenCode and others) and asks where to install. Add `--global` for a user-wide install, or `--yes` for unattended installation. The [skill](skills/jevgrep/SKILL.md) explains installation, invocation and the meaning of returned context. It leaves research and implementation decisions to the calling agent. The current repository skill checks for `jg` and installs the CLI if it is missing; authentication still needs your selected provider’s key. The skill installer itself does not configure credentials.

`jg skill` delegates to the [skills CLI](https://github.com/vercel-labs/skills) and needs npm/npx plus network access. You can also run that installer directly, without the CLI installed:

npx skills add dzhng/jevgrep --skill jevgrep

In 0.1.0, `jg skill` only prints the bundled skill; use `npx skills` with that version.

Upgrade

There is currently no `jg upgrade` command. Upgrade the CLI with npm:

npm install -g @dzhng/jevgrep@latest
jg --version

Update the installed skill separately by rerunning `jg skill`. Updating the npm package does not overwrite skill files in your projects. See the [package guide](apps/cli/README.md) for authentication details.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 180 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. ![jevgrep — Same intelligence. About 30% lower cost.](assets/cover.png)
  2. # jevgrep
  3. [![npm](https://img.shields.io/npm/v/@dzhng/jevgrep?style=flat-square&color=ef5638)](https://www.npmjs.com/package/@dzhng/jevgrep)
  4. [![MIT license](https://img.shields.io/badge/license-MIT-blue?style=flat-square)](LICENSE)
  5. [![Node.js 22+](https://img.shields.io/badge/Node.js-22%2B-339933?style=flat-square)](apps/cli/README.md)
  6. [![Release](https://img.shields.io/github/actions/workflow/status/dzhng/jevgrep/publish.yml?style=flat-square&label=release)](https://github.com/dzhng/jevgrep/actions/workflows/publish.yml)
  7. **Same intelligence. ~30% lower cost.**
  8. Find code by asking what it does. In our ten-task SWE-bench comparison, Jevgrep
  9. successfully completed the same 8 of 10 tasks as the baseline, at lower cost.
  10. Coding agents spend part of every unfamiliar task finding the right files.
  11. Jevgrep gives them a place to start: ask a repository question, and `jg` returns
  12. relevant files, reading leads, and verbatim source excerpts in one stdout response.
  13. It uses [Jev](https://vercel.com/ai-gateway/models/jev) to judge relevance across
  14. folders, files, and declarations. Your coding agent then implements and tests the change.
  15. ```sh
  16. npm install -g @dzhng/jevgrep
  17. jg auth
  18. jg skill
  19. jg "How are telemetry events recorded and sent?" ./my-project
  20. ```
  21. Requires **Node.js 22+**, **macOS, Linux, or Windows**, and a key for **Vercel AI Gateway, TypeSafe, OpenRouter, OpenCode Zen, or a custom TypeSafe-compatible endpoint**.
  22. No separate Python, Bun, or ripgrep installation is required to use `jg`.
  23. ## Install the agent skill — required for agent setup
  24. Installing the CLI alone does not teach your coding agent to use it. **Install
  25. the skill as well**, from the project where your agent works:
  26. ```sh
  27. jg skill
  28. ```
  29. The installer detects your coding agents (Claude Code, Codex, OpenCode and
  30. others) and asks where to install. Add `--global` for a user-wide install, or
  31. `--yes` for unattended installation. The
  32. [skill](skills/jevgrep/SKILL.md) explains installation, invocation and the meaning of returned context.
  33. It leaves research and implementation decisions to the calling agent. The current repository skill
  34. checks for `jg` and installs the CLI if it is missing; authentication still needs
  35. your selected provider’s key. The skill installer itself does not configure credentials.
  36. `jg skill` installs the `jevgrep` skill from the repository's `main` branch using
  37. a pinned [skills CLI](https://github.com/vercel-labs/skills), with npm lifecycle
  38. scripts disabled. Skill installation needs Node 22.20+ and npm/npx plus network
  39. access for the installer. Rerun `jg skill` to get the current skill without
  40. waiting for a new npm release.
  41. ### Upgrade
  42. There is currently no `jg upgrade` command. Upgrade the CLI with npm:
  43. ```sh
  44. npm install -g @dzhng/jevgrep@latest
  45. jg --version
  46. ```
  47. After updating the CLI, refresh the installed skill by rerunning `jg skill`.
  48. Updating the npm package does not overwrite skill files in your projects. The
  49. installer records the GitHub source and skill path in `skills-lock.json`; rerun
  50. `jg skill` on another machine or after changing your Node installation. See the
  51. [package guide](apps/cli/README.md) for authentication details.
  52. ## Start with a question, leave with source
  53. Use `jg` when you know the behavior you need to understand but not where it lives:
  54. ```sh
  55. jg "Where is authentication checked before a request reaches a handler?" .
  56. jg "How are database connections created, pooled, and closed?" ./src
  57. jg "Which tests cover retry behavior when a request times out?" .
  58. ```
  59. Jevgrep explores the repository hierarchy and follows qualifying branches. It
  60. selects files using content previews, then identifies useful source units and
  61. surrounding context. It keeps qualifying file locations even when it cannot
  62. confidently return an excerpt; it does not force every search into a fixed top-two
  63. list.
  64. The summary and compact file list come first, followed by selected source with
  65. line references, then detailed declaration and call locations. Python, TypeScript/JavaScript, Go and Rust support declaration
  66. parsing, with extensionless Python, Node, Bun and Deno scripts recognised by their shebang;
  67. other text uses a fallback. The output is evidence for the agent to use,
  68. not a generated answer or a guarantee that every relevant file was found.
  69. [See a recorded output example](specs/done/jevgrep/assets/stdout-example.txt).
  70. When you already know an exact symbol or path, a direct read or `rg` search may be
  71. all you need. Jevgrep is most useful for questions that span unfamiliar files.
  72. ## What we measured
  73. **Same intelligence, ~30% lower coding-agent cost.** Both
  74. Jevgrep and the no-Jev baseline solved **8/10 tasks**. Full Sol cost fell from
  75. **$7.62 to $5.44**—a measured **28.6% reduction**, rounded to ~30%—including failed
  76. attempts and excluding Jev cost.
  77. ![Jevgrep workflow: about 30% lower coding-agent cost, with 8 of 10 tasks solved both with and without Jevgrep.](assets/how-it-works.png)
  78. This comparison uses ten tuned Python SWE-bench tasks, one frozen installed
  79. package and the exact public skill in this repository. It measures task success
  80. and cost, not a speed improvement or guaranteed savings on every repository.
  81. See the [results and methodology](evals/results/relevance-threshold-2026-09-27.md)
  82. for per-task costs, artifact identities and limitations. A separate
  83. [speed study](evals/results/speed-2026-09-28.md) measures the follow-up local
  84. optimizations with Jev’s native TypeSafe endpoint.
  85. The [0.4.3 total-cost rerun](evals/results/total-cost-2026-09-28.md), including
  86. Jev, measured **25.8% lower total cost with the same 8/10 tasks solved**.
  87. The older ~30% graphic above reports Sol-only cost. Future benchmark totals include Jev.
  88. The [0.5.0 evaluation](evals/results/combined-cost-research-2026-09-28.md) retained
  89. 8/10 solves while reducing native Jev cost by about 59% versus that 0.4.3 run.
  90. Combined Sol-plus-Jev cost was 2–3% higher, accepted as a small tradeoff for this
  91. release. These single-run observations do not establish statistical equivalence
  92. or a speed improvement.
  93. Current protocols and subsequent experiments are in the
  94. [evaluation records](evals/README.md).
  95. ## Source, credentials, and local state
  96. Searches send eligible source content to Jev through the provider selected during auth. Default
  97. filesystem filtering respects ignore files and excludes hidden, dependency/build,
  98. binary, and obvious credential files, including `.env` variants and any path
  99. component whose name ends in `.env`.
  100. It also excludes a whole file as `sensitive_content` when its text matches a
  101. conservative credential heuristic: a known provider token prefix (such as `npm_`
  102. or `glpat-`), a Slack webhook URL, or a credential-shaped assignment with a
  103. long, high-entropy value. UUID-shaped identifiers are excluded from that entropy
  104. signal. This is best-effort detection, not redaction or a guarantee that all
  105. sensitive information has been removed; choose a search root you intend to send.
  106. Use `--include-sensitive` when a search must include sensitive names, private-key
  107. markers, and content-level matches.
  108. `jg files [root]` counts the files a search under that root may read, grouped by
  109. top-level directory, with no provider key or network request. It takes the same
  110. filtering flags as search. Use `--only code,config,docs` or
  111. `--exclude-types data,logs,temp` to narrow the scope by filename category before
  112. paid evaluation. See [choosing a search scope](docs/file-types.md) for previews
  113. and the limits of cost savings.
  114. To skip paths inside that root for one search, pass `--exclude` with a gitignore pattern
  115. relative to the root, for example `--exclude '**/*.test.ts' --exclude 'src/generated/'`.
  116. `jg auth` asks for your provider, then saves its key in an owner-only config file.
  117. Re-running auth replaces that setup; searches always use the saved provider.
  118. `jg doctor` checks it with synthetic input. Existing saved keys without a provider
  119. remain Vercel keys. Environment-based credentials and endpoint overrides are not
  120. used; run `jg auth` if you previously relied on them.
  121. Evaluation answers are cached locally by default. The CLI writes its output to
  122. stdout and does not create report files. Use `jg --help` for cache controls,
  123. search overrides, and incomplete-result behavior.
  124. ## Development
  125. The repository uses TypeScript, Bun workspaces, and Turborepo. From a checkout:
  126. ```sh
  127. bun install --frozen-lockfile
  128. bun run dev --help
  129. bun run verify
  130. ```
  131. Verification includes Docker tests of the installed Node-only package. For the
  132. reasoning behind retrieval, parsing, caching, and failure handling, start with the
  133. [architecture](docs/architecture.md) and [implementation record](specs/done/jevgrep/README.md).
  134. [Release guidance](scripts/RELEASING.md) covers tag-triggered npm publication and
  135. verification of the exact public package.
  136. See [Contributing](CONTRIBUTING.md) for the pull-request review and maintenance
  137. policy.
  138. [MIT](LICENSE). [Artwork and generation prompts](assets/README.md).
  139. ## Ratchet Files
  140. Source files are limited to 8,000 estimated tokens (bytes divided by four). The pre-commit hook runs `scripts/check-file-size.mjs` through lint-staged. `bun install` runs `scripts/install-git-hooks.mjs`, which installs the hook once in the repository's shared hooks directory, so every linked worktree runs it; each checkout that commits needs its own `bun install`, and commits from one without dependencies are refused. Installing a branch from before this change re-enables Husky for every checkout until a current branch runs `bun install` again. Existing oversized authored files are frozen at their current size; generated and vendored paths are excluded through `.gitattributes`. A file may use `@ratchet-files-ignore` in its first 10 lines only when splitting it would cost more than it saves, with the reason beside the pragma.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.