What is JevTrace and how does it help coding agents?

An MCP server that returns the JS/TS code a given task needs within a token budget. It uses the TypeScript compiler to follow calls, types, callers and tests, and Jev to judge which of those candidates are actually relevant.

What does its benchmark claim?

On 19 JS/TS tasks it delivered 98.2% of required code versus jevgrep's 70.6%, with 5.9× lower Jev cost, 3.0× faster search and 53% fewer tokens handed to the agent. The labels were written by the JevTrace authors.

How does Jev fit into the pipeline?

Compiler analysis supplies candidates; Jev decides relevance with typed judgments, so results are ranked rather than dumped. You can route Jev through OpenRouter, TypeSafe, Vercel AI Gateway, OpenCode Zen or your own endpoint.

Run npx -y jevtrace setup once to pick a provider and enter its API key, then add it to a client: claude mcp add jevtrace --scope user -- npx -y jevtrace. Codex and Cursor take one config block each, documented in docs/clients.md.

What kinds of projects does it support?

TypeScript and JavaScript — .ts, .tsx, .js, .jsx, .mjs, .cjs — plus Vue components, ESM and CommonJS, with or without a tsconfig, up to 20,000 source files. Svelte and Astro are not supported. Node.js 20+ is required.

Can I try it without a coding agent?

Yes: npx -y jevtrace query --root /path/to/your/project --task "Fix refresh token validation" runs a retrieval directly. Ask about one behaviour or area; project-wide requests are met with per-area subtasks instead of code.

What are the limitations to keep in mind?

It answers 'where is X decided and what touches it' questions about existing code — not open-ended reviews. The headline numbers are the authors' own retrieval benchmark with self-written labels, and the captured README is brief.

Jev directory

CommunityJev Agents & MCP

JevTrace

An MCP server that gives coding agents only the JS/TS code a task needs: the TypeScript compiler follows calls, types, callers and tests, Jev judges relevance, and results come back within a token budget.

See highlights Open github.com

Stars: 3
Language: TypeScript
License: MIT
Reading time: 2 min
Published by: Community
Author: 123wwwa
Last verified: Oct 2026

123wwwa/jevtrace

User repository on GitHub

View on GitHub

An MCP server that cuts coding-agent token usage by retrieving only the relevant JS/TS code using the TypeScript compiler and Jev.

Forks: 0
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.Homepage

Tagscommunity catalog-import logicrw mcp code-search

Works forJev for Code Jev for Tool calling

Used inJev for Software & DevTools

Languages: TypeScript
Platforms: MCP CLI OpenRouter Vercel AI Gateway
Capabilities: Open sourceRequires API key

Highlights

On the authors' 19-task benchmark it delivered 98.2% of required code versus jevgrep's 70.6%, at 5.9× lower Jev cost and 53% fewer tokens.
Compiler analysis follows calls, types, callers and tests; Jev judges relevance via OpenRouter, TypeSafe, Vercel AI Gateway or your own endpoint.
Two commands set it up: npx -y jevtrace setup, then claude mcp add jevtrace --scope user -- npx -y jevtrace.
Handles .ts/.tsx/.js/.jsx/.mjs/.cjs and Vue components, ESM and CommonJS, with or without tsconfig, up to 20,000 source files.
Requires Node.js 20+ and is MIT-licensed; project-wide requests like 'find bugs' get per-area subtasks to call it with instead of code.

Quickstart

bash

npx -y jevtrace setup
claude mcp add jevtrace --scope user -- npx -y jevtrace

Watch out

Copy reviewed against the repo README on 2026-10-15; the capture is short and the benchmark method lives in linked docs. The headline figures come from the authors' own retrieval benchmark with labels they wrote — per-task results and limitations are in docs/benchmark-vs-jevgrep.md.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

X: Jev in front of the HeyGen MCP

X: Playwright CLI + Jev vs Playwright MCP

X: A local MCP server that cuts computer use from 23s to 0.8s. Cartwright built Cortex, a local MCP server where…

X: A side panel that drives any site's WebMCP tools

From the Jev guides

Original write-ups on what this entry is for.

MCP tool routing: Jev in front of the tool call

Oct 16, 2026

Tool menus can outgrow the context that reads them. The catalog's Jev-based routers and gates narrow the list: benchmarked tool search, per-tool scoring, fail-closed approvals, and WebMCP in the browser.

Jev Agents & MCP#mcp#tool-routing#agents

Read article

Approval policies for coding agents: hooks, guards, and gateways

Oct 16, 2026

How the Jev guard family turns coding-agent permissions into auto-approve, deny, or ask: hooks that enforce, guards that fail safe, and gateways that gate generation.

Jev Agents & MCP#coding-agents#guardrails#hooks