When do I use judge versus rerank versus systemone?

Judge compiles natural language into a typed TypeSafe request using the fleet Ollama model with fallbacks, then executes it. Rerank scores each item with a comparable Score; systemone runs your raw state and questions directly.

Does every call depend on the local Ollama fleet?

No. Judge and rerank compile through qwen3.8:27b with fallback and wake support, with robust think:false stripping and JSON repair. Raw systemone never touches the fleet.

How do I register it without leaking my API key?

Export TYPESAFE_API_KEY or keep it in ~/typesafe-mcp/.env, which stays gitignored. Fleet registration carries no secret, and the key is never written to disk image output.

Can I test the logic without running MCP plumbing?

Core logic lives in core.py with no mcp dependency for isolated testing, and server.py is a thin shim supporting both FastMCP v1 and MCPServer v2 plus optional chain and wake env vars.

What does the systemone tool do compared with judge and rerank?

systemone is a raw passthrough: you supply a state and questions and it executes them with no fleet LLM. judge compiles natural language into a typed TypeSafe request, and rerank orders a list by a criterion using a comparable Score per item.

Which model work does the Ollama fleet perform?

judge and rerank use the fleet, qwen3.8:27b at corsair with a fallback chain and wake via ollawake, to compile the request. systemone does not touch the fleet at all.

How is the TypeSafe key kept out of the MCP registration?

The key is read from the TYPESAFE_API_KEY environment variable or, if absent, from ~/typesafe-mcp/.env, so the openclaw and hermes registration does not carry the secret. The .env file is gitignored, and the key is never written to disk or the image.

CommunityJev Agents & MCP

TypeSafe MCP — Obrais-cloud

Python MCP server over stdio that exposes Jev to an agent fleet with judge, rerank, and raw systemone tools plus Ollama-backed compilation.

See highlights

Open github.com

Stars: 1
Language: Python
Published by: Community
Author: Obrais-cloud
Last verified: Oct 2026

Obrais-cloud/typesafe-mcp

User repository on GitHub

View on GitHub

MCP server exposing TypeSafe (Jev/System One) to the fleet: judge, rerank, systemone

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

This is one of several Jev listings for the same project. The canonical entry is Typesafe MCP — itsmostafa.

About this project

MCP server que expone TypeSafe (Jev / System One) a la flota (openclaw + hermes). judge/rerank usan la flota Ollama para compilar (qwen3.8:27b @ corsair, con fallback y wake vía ollawake). systemone no toca la flota. Env opcionales: TSMCPLLMCHAIN (JSON de la cadena de fallback), TSMCPOLLAWAKE. core.py — lógica (TypeSafe + compilador + fallback de flota + rerank), sin dependencia de mcp; testeable sola. server.py — wrapper MCP fino. Shim compatible: mcp v1 (FastMCP) y v2 (MCPServer).

Quoted from the project's README. Read the full README.

Tagscommunity github

Works forJev for Tool calling Jev for Ranking

Used inJev for Software & DevTools

Languages: Python

Highlights

Judge turns natural language into typed TypeSafe requests and returns request plus probabilities.
Rerank orders id-text lists by a natural-language criterion using per-item comparable Scores.
Raw systemone passthrough executes caller-supplied state and questions with no fleet LLM.
Stdio server with v1/v2 shim, secret-by-env design, and separately testable core module.

Watch out

Fleet-dependent setup needing Ollama and wake tooling; raw systemone bypasses the fleet LLM entirely. Copy reviewed against the repository README on 2026-10-04, not from running it.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

jev-decision-maker-at-meteora

Condor drop-in where Jev Noul fan-out ranks Meteora pools and a Score sizes positions, while math clamps enforce rug cards, cost tiers, dry-run, and one-sided walls.

Jev Agents & MCP#community#github

Community

cedarmuse-creator/jev-decision-maker-at-meteoraPython

jev-judge-mcp

MCP server over stdio that gives coding agents fourteen Jev judgment tools, turning evidence plus an enumerable question into auto, review, or escalate actions from calibrated probabilities.

Jev Agents & MCP#community#jevbest#catalog-import

Community

PyModel/jev-judge-mcpPython

jev-reranking

Reproducible study testing whether Jev zero-shot can match specialist rerankers on fixed MS MARCO passage, MS MARCO document, and TREC-1 WSJ candidates, reporting ranking quality, time, and API cost.

Jev Agents & MCP#community#github

Community

carlaiau/jev-rerankingPython

hermes-jev

Async supervisory layer for Hermes Kanban workers: locked DoD contracts, evidence tracking, budgeted trajectory checks, and Jev only for unresolved semantic criteria.

Jev Agents & MCP#community#jevbest#catalog-import

Community

keeltrace/hermes-jevPython

langchain-skill-router

LangChain deepagents middleware that replaces SkillsMiddleware with per-turn judging, loading only the SKILL.md files a turn needs so large catalogs stay out of the prompt.

Jev Agents & MCP#community#jevbest#catalog-import

Community

deyna256/langchain-skill-routerPython

typesafe-cli

Stdlib-only Python CLI and agent skill that evaluates a state against typed Choice, Score and Noul questions over the System One API without generating prose.

Jev Agents & MCP#community#jevbest#catalog-import

Community

geilt/typesafe-cliPython

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai