jevgrep

Same intelligence. ~30% lower cost.

Find code by asking what it does. In our ten-task SWE-bench comparison, Jevgrep successfully completed the same 8 of 10 tasks as the baseline, at lower cost.

Coding agents spend part of every unfamiliar task finding the right files. Jevgrep gives them a place to start: ask a repository question, and jg returns relevant files, reading leads, and verbatim source excerpts in one stdout response. It uses Jev to judge relevance across folders, files, and declarations. Your coding agent then implements and tests the change.

npm install -g @dzhng/jevgrep
jg auth
jg skill
jg "How are telemetry events recorded and sent?" ./my-project

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. No separate Python, Bun, or ripgrep installation is required to use jg.

Install the agent skill — required for agent setup

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 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 installs the jevgrep skill from the repository's main branch using a pinned skills CLI, with npm lifecycle scripts disabled. Skill installation needs Node 22.20+ and npm/npx plus network access for the installer. Rerun jg skill to get the current skill without waiting for a new npm release.

Upgrade

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

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

After updating the CLI, refresh the installed skill by rerunning jg skill. Updating the npm package does not overwrite skill files in your projects. The installer records the GitHub source and skill path in skills-lock.json; rerun jg skill on another machine or after changing your Node installation. See the package guide for authentication details.

Start with a question, leave with source

Use jg when you know the behavior you need to understand but not where it lives:

jg "Where is authentication checked before a request reaches a handler?" .
jg "How are database connections created, pooled, and closed?" ./src
jg "Which tests cover retry behavior when a request times out?" .

Jevgrep explores the repository hierarchy and follows qualifying branches. It selects files using content previews, then identifies useful source units and surrounding context. It keeps qualifying file locations even when it cannot confidently return an excerpt; it does not force every search into a fixed top-two list.

The summary and compact file list come first, followed by selected source with line references, then detailed declaration and call locations. Python, TypeScript/JavaScript, Go and Rust support declaration parsing, with extensionless Python and Node scripts recognised by their shebang; other text uses a fallback. The output is evidence for the agent to use, not a generated answer or a guarantee that every relevant file was found. See a recorded output example.

When you already know an exact symbol or path, a direct read or rg search may be all you need. Jevgrep is most useful for questions that span unfamiliar files.

What we measured

Same intelligence, ~30% lower coding-agent cost. Both Jevgrep and the no-Jev baseline solved 8/10 tasks. Full Sol cost fell from $7.62 to $5.44—a measured 28.6% reduction, rounded to ~30%—including failed attempts and excluding Jev cost.

This comparison uses ten tuned Python SWE-bench tasks, one frozen installed package and the exact public skill in this repository. It measures task success and cost, not a speed improvement or guaranteed savings on every repository. See the results and methodology for per-task costs, artifact identities and limitations. A separate speed study measures the follow-up local optimizations with Jev’s native TypeSafe endpoint.

The 0.4.3 total-cost rerun, including Jev, measured 25.8% lower total cost with the same 8/10 tasks solved. The older ~30% graphic above reports Sol-only cost. Future benchmark totals include Jev.

The 0.5.0 evaluation retained 8/10 solves while reducing native Jev cost by about 59% versus that 0.4.3 run. Combined Sol-plus-Jev cost was 2–3% higher, accepted as a small tradeoff for this release. These single-run observations do not establish statistical equivalence or a speed improvement.

Current protocols and subsequent experiments are in the evaluation records.

Source, credentials, and local state

Searches send eligible source content to Jev through the provider selected during auth. Default filesystem filtering respects ignore files and excludes hidden, dependency/build, binary, and obvious credential files, including .env variants and any path component whose name ends in .env. It also excludes a whole file as sensitive_content when its text matches a conservative credential heuristic: a known provider token prefix (such as npm_ or glpat-), a Slack webhook URL, or a credential-shaped assignment with a long, high-entropy value. UUID-shaped identifiers are excluded from that entropy signal. This is best-effort detection, not redaction or a guarantee that all sensitive information has been removed; choose a search root you intend to send. Use --include-sensitive when a search must include sensitive names, private-key markers, and content-level matches. jg files [root] counts the files a search under that root may read, grouped by top-level directory, with no provider key or network request. It takes the same filtering flags as search. To skip paths inside that root for one search, pass --exclude with a gitignore pattern relative to the root, for example --exclude '**/*.test.ts' --exclude 'src/generated/'.

jg auth asks for your provider, then saves its key in an owner-only config file. Re-running auth replaces that setup; searches always use the saved provider. jg doctor checks it with synthetic input. Existing saved keys without a provider remain Vercel keys. Environment-based credentials and endpoint overrides are not used; run jg auth if you previously relied on them. Evaluation answers are cached locally by default. The CLI writes its output to stdout and does not create report files. Use jg --help for cache controls, search overrides, and incomplete-result behavior.

Development

The repository uses TypeScript, Bun workspaces, and Turborepo. From a checkout:

bun install --frozen-lockfile
bun run dev --help
bun run verify

Verification includes Docker tests of the installed Node-only package. For the reasoning behind retrieval, parsing, caching, and failure handling, start with the architecture and implementation record. Release guidance covers tag-triggered npm publication and verification of the exact public package.

See Contributing for the pull-request review and maintenance policy.

MIT. Artwork and generation prompts.

Ratchet Files

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, so a clone must install the repository dependencies before commits. 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.