What is jevcrawl and what does it do?

An MCP server and CLI web crawler for AI agents: a small decision model picks every link and every passage, so the agent's LLM reads only the pages that answer its question instead of whole sites.

Does jevcrawl actually use Jev or TypeSafe models?

The README excerpt names "a small decision model" that picks links and passages but does not mention Jev, TypeSafe, or System One; the jev in its name and the decision-model premise are the only signals in the digest, so treat the Jev link as a class resemblance.

How is jevcrawl different from a normal crawler?

Instead of reading everything a crawler fetches or letting the LLM pick each link, jevcrawl uses a decision model to choose links and passages; the README claims it finds the answer page 2x more often than Crawl4AI's goal-directed crawl.

What do the WebWalkerQA numbers say?

On 50 tasks: 40% answered correctly and the answer page found 54% of the time for about $0.025 per task, against 30% and $0.194 when Claude picks every link and 8% and $0.228 for a breadth-first crawl.

How do I install or run jevcrawl?

It ships on PyPI as jevcrawl, supports Python 3.11 and newer, and runs as an MCP server or CLI, per the badges and the README's Quick start heading; the digest truncates before the commands themselves.

What does the jevcrawl demo show?

Crawling docs.python.org for what the walrus operator := does: 6 pages, 80,034 tokens reduced to 1,873, $0.0069, and 8.3 seconds, returning the What's New In Python 3.8 passage, according to the README.

What license does jevcrawl use and how active is it?

Apache-2.0 with CI on GitHub Actions; metadata in the digest shows 77 stars, 6 forks, Python, created 2026-10-11, and a last push on 2026-10-15.

Jev directory

CommunityJev Agents & MCP

Jevcrawl

An MCP server and CLI web crawler for AI agents in which a small decision model picks every link and passage, so the agent's LLM reads only what answers its question. On 50 WebWalkerQA tasks the README claims 40% correct answers at about $0.025 per task.

See highlights Open github.com

Stars: 77
Language: Python
License: APACHE-2.0
Reading time: 3 min
Published by: Community
Author: RayenMlayeh
Last verified: Oct 2026

rayenmlayeh/jevcrawl

User repository on GitHub

View on GitHub

MCP server and CLI that crawls websites for AI agents: finds the answer page 2x more often than Crawl4AI, 10x cheaper per answer than letting Claude crawl.

Forks: 6
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.

Tagscommunity github python mcp crawler

Works forJev for Tool calling Jev for Ranking

Used inJev for Software & DevTools

Languages: Python
Platforms: MCP CLI
Capabilities: Open source

Highlights

README: a small decision model picks every link and every passage, so your agent's LLM reads only what answers its question, at a few cents per crawl.
On 50 WebWalkerQA tasks the project answered 40% correctly and found the answer page 54% of the time for $0.025 per task.
The same scoreboard compared Claude picking every link (30%, $0.194 per task) and a breadth-first crawl (8%, $0.228).
Demo: a walrus-operator question on docs.python.org answered in 8.3 seconds after 6 pages and 80,034 tokens were reduced to 1,873 for $0.0069.
Apache-2.0, Python 3.11 and newer, with CI; GitHub metadata shows 77 stars, 6 forks, created 2026-10-11, pushed 2026-10-15.

Watch out

README reviewed 2026-10-17 and truncated inside the "Why jevcrawl" bullets, so install commands, configuration, the tools list, and how the decision model is served are not in the digest. The 40% and 54% figures and every cost number are the project's own 50-task WebWalkerQA benchmark, not independent results.

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