User repository on GitHub
System One MCP connector to evaluate anything fast and cheap. Give your AI agent direct access to models like: Jev, D1, CLM and Laya
- Forks
- 41
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Tools & CLIs
Typesafe MCP — itsmostafa
A single-binary Go MCP server exposing one evaluate tool for noul, choice, and score questions, with one-command registration for Claude Code, Claude Desktop, and Codex.
- Stars
- 353
- Language
- Go
- License
- MIT
- Reading time
- 3 min
- Published by
- Community
- Author
- itsmostafa
- Last verified
- Sep 2026
About this project
Give your AI agent direct access to models like Jev, CLM, Laya, and d1: quick decisions with real probabilities. evaluate connects Claude Code, Claude Desktop, Codex, Hermes and pi to TypeSafe's Jev model. Your agent asks a question like "is this urgent?" or "which team owns this?" and gets back a number or an option it can use in an if statement.
Quoted from the project's README. Read the full README.
Highlights
- Exposes one MCP tool, evaluate, with noul, choice, and score question types.
- One command registers Claude Code, Claude Desktop, and Codex when each is detected.
- Retries 429 and 529 with exponential backoff, with a 60-second timeout and 16 MiB response cap.
- Ships as a single static Go binary with a checksum-verified self-update.
Watch out
Needs a TypeSafe or OpenRouter key, and OpenRouter's decisions endpoint is still alpha and may move.
Questions about this entry
Short answers drawn from the linked project or page.
Effectively, yes: a single static Go binary exposing one MCP tool named evaluate, with noul, choice and score question types scored in parallel in one call. What it adds is the plumbing you would otherwise write - client registration, retries on 429 and 529 with backoff, a 60-second timeout, a 16 MiB response cap and a checksum-verified self-update.
Up to 500 records at once by passing them as items and asking the same questions of each; if one record fails, the rest still complete. Unrelated judgments travel together too, so urgency, ownership and sentiment can be asked in a single request.
Yes. Point TYPESAFE_BASE_URL at your server and keep TYPESAFE_API_KEY set, since it selects the route; any non-empty value works if your server ignores keys. CLM needs TYPESAFE_MODEL=clm-latest, Liquid AI's d1 needs their hosted endpoint, and saved profiles let you switch without rerunning setup.
The quickstart documents macOS and Linux only, and nothing in the README covers Windows. Installing is a curl-piped install.sh, so on Windows you would be running a shell script that is not documented as supported.
Install with the curl script from the README, then run TYPESAFE_API_KEY=your-key evaluate setup mcp. That finds Claude Code, Claude Desktop, Codex and Hermes and registers evaluate with each, and running it again updates them. For pi the command is evaluate setup pi, and you can supply OPENROUTER_API_KEY instead if you already have OpenRouter.
Any model behind the System One API. Documented setups cover Liquid AI's d1 on their hosted endpoint, Jared Palmer's Kev 4B and Respan's Span-01 through OpenRouter profiles, and self-hosted CLM or Laya. Span-01 supports noul questions only, with state as text or a conversation. Save each as a profile and switch with evaluate profile use.
It retries 429 and 529 responses with exponential backoff, applies a 60-second timeout, and caps a response at 16 MiB. Datasets pass up to 500 records as items in one call, and if a single record fails the rest still complete. It ships as a single static Go binary with a checksum-verified self-update.
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
Other listings of this project
Forks and copies of this project in the directory. This entry is the canonical one.
Python MCP server over stdio that exposes Jev to an agent fleet with judge, rerank, and raw systemone tools plus Ollama-backed compilation.
Jev Agents & MCP#community#github
Community
Obrais-cloud/typesafe-mcpPython
Obrais-cloud/typesafe-mcpPythonA Model Context Protocol server that exposes TypeSafe AI's Jev decision model as typed tools over stdio, runnable with npx or bunx. Ships jev_decide, jev_choice, jev_score, jev_noul, jev_box and jev_list_models for MCP clients such as Claude Code, Cursor and VS Code.
Jev Agents & MCP#mcp#typescript#agents
Communitydevbux/typesafe-mcp
More Jev resources like this
A Rust CLI and stdio MCP server that grounds repository and crate discovery in live GitHub and crates.io metadata, then lets Jev score candidates so only verified, installable packages are returned.
Jev Tools & CLIs#community#rust#cli
Community
AkashPriyadarshii/jev-scoutRust
AkashPriyadarshii/jev-scoutRustAn unofficial Python CLI and stdio MCP server for Jev: send text or JSON state with typed questions and get machine-readable noul, choice, or score answers from the official API, OpenRouter, Vercel, or a proxy.
Jev Tools & CLIs#community#python#cli
Community
tumf/jev-cliPython
tumf/jev-cliPythonAn open-source, local-first memory layer that captures session history across Claude Code, Cursor, Codex, OpenCode, Cline, and 20+ other harnesses, then has Jev judge bounded, redacted trace projections for reusable lessons a person reviews.
Jev Tools & CLIs#community#go#agents
Community
Asymptote-Labs/agent-beaconGo
Asymptote-Labs/agent-beaconGoAn MCP server and agent skill that verifies coding-agent work by driving the real running app and returning pass, fail, or couldn't-tell verdicts with the file:line to fix; its listing frames the approach as Jev-style machine-native runtime perception.
Jev Tools & CLIs#community#typescript#mcp
Community
reticlehq/reticleTypeScript
reticlehq/reticleTypeScriptA browser-use agent where Jev picks one action per step from a page's interactive elements. Runs as an MCP server, CLI, or library and returns a step trace, screenshot, and console errors.
Jev Tools & CLIs#community#typescript#mcp
Community
jkudish/jev-browserJavaScript
jkudish/jev-browserJavaScriptAn MCP server exposing ten Jev judgment tools, including verify, screen, find, rerank, classify, review, and gate, each returning typed probabilities in roughly 150 to 500 ms.
Jev Tools & CLIs#community#typescript#mcp
Communityjkudish/jev-mcpJavaScript
jkudish/jev-mcpJavaScriptFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.