What is jevreel and who builds it?

A zero-embedding video intelligence project from DeepAleph, shipped as MCP tools and a skill: it lays a video out on labelled contact sheets and has Jev judge every sheet in parallel.

How does jevreel use Jev?

Jev, TypeSafe's fast decision model, judges each contact sheet in parallel; the best windows are re-checked on full frames, and the agent receives moments, times and probabilities rather than generated text.

How do I connect jevreel to my agent?

It runs as an MCP server plus a skill for Claude Code, Codex, Cursor, Claude Desktop or anything that speaks MCP. The digest was truncated before the install section, so exact setup commands were not captured.

What does a jevreel query cost?

The README's five-session demo lists 8 to 19 Jev requests per question and $0.0003 to $0.0064 per query. A Jev API key is required, so normal TypeSafe API charges apply.

Does jevreel store embeddings or vectors?

No. The README states that no vectors are computed or stored anywhere; video is represented as contact sheets, and frames are re-checked on demand when a window needs verification.

What video evidence does the README show?

Five unedited Claude Code sessions against Blender's open movie Tears of Steel (12:14, CC BY 3.0), each clip drawn from its recording with the tool calls, Jev answers, hits and cost shown.

What could not be verified from the captured README?

The capture ends inside the fifth demo row, so install steps, configuration and the sections after the demo were not available; treat the demo numbers as the author's own runs, not a published benchmark.

Jev directory

CommunityJev Agents & MCP

Jevreel

A zero-embedding approach to video intelligence from DeepAleph: jevreel lays a video out on labelled contact sheets, one frame per second and sixty frames to an image, and has Jev judge every sheet in parallel. Your coding agent gets moments, times and probabilities back over MCP.

See highlights Open github.com

Stars: 24
Language: Python
License: MIT
Reading time: 2 min
Published by: Community
Author: DeepAleph
Last verified: Oct 2026

deepaleph/jevreel

Organization repository on GitHub

View on GitHub

A zero-embedding approach to video intelligence, using Jev: contact sheets + typed decisions, as MCP tools and a skill for any coding agent.

Forks: 2
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.

Tagscommunity github mcp video python

Works forJev for Extraction Jev for Scoring Jev for Ranking

Used inJev for Media & Entertainment

Languages: Python
Platforms: MCP
Capabilities: Open sourceRequires API key

Highlights

Lays a video out on labelled contact sheets, one frame per second and sixty frames to an image, then has Jev judge every sheet in parallel.
Best windows are re-checked on full frames, and the agent gets back moments, times and probabilities; no vectors are computed or stored.
Works as MCP tools and a skill for Claude Code, Codex, Cursor, Claude Desktop or anything that speaks MCP.
Demo: a 12-minute film becomes 13 contact sheets; five unedited Claude Code sessions report hits, costs and quoted answers on Blender's Tears of Steel.
Demo cost per query ranges from $0.0003 in speech mode to $0.0064 in search by photo, at 8 to 19 Jev requests each.

Watch out

README reviewed 2026-10-17; the digest truncates the demo session table after the fourth row and cuts off the fifth, so the install command and the rest of the demo evidence were not available. All costs, hit times and probabilities are the author's own single-film runs on Tears of Steel, not an independent benchmark.

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