Releases·Tool Release
Unofficial community skill teaches coding agents to build with TypeSafe Jev's typed decisions and calibrated confidence
A community-published GitHub skill teaches coding agents to build with TypeSafe AI's Jev, a System One model that answers typed questions with values and calibrated probabilities in 100-200 ms. The skill is unofficial and not endorsed by TypeSafe AI, which publishes its own skill separately. It ships with prior art from 150+ community projects sorted by pattern shape.
98 engagement1 sourceSat, Sep 26, 2026, 12:01 AM
AI Overview
- Unofficial GitHub skill (98 engagement) teaches coding agents to use Jev for typed decisions with calibrated confidence, distinct from TypeSafe AI's official skill
- Jev is a System One model that answers typed questions with a value and calibrated probability, usually in 100-200 ms
- Skill links to 150+ community projects sorted by shape, each with a code sketch, and installs in Claude Code, Codex, Antigravity CLI, Muse, and Muse Code
- TypeSafe publishes its own official skill at typesafe-ai/skills; the community skill is unaffiliated and unendorsed
Related
Social·Community
awesome-jev: curated list of projects built on TypeSafe AI's Jev typed-decision model
A curated GitHub repository, awesome-jev, aggregates public projects, integrations, and discussions built on Jev, TypeSafe AI's System One model for typed decisions. Jev is not a chat model: it takes unstructured state plus a typed question and returns a typed decision — a choice, score, or boolean with confidence — positioning it as a drop-in decision layer for classification, routing, scoring, verification, and agent guardrails. The list gives practitioners a fast-scanning field guide to which decision patterns transfer across industries.
A curated GitHub repository, awesome-jev, aggregates public projects, integrations, and discussions built on Jev, TypeSafe AI's System One model for typed decisions. Jev is not a chat model: it takes unstructured state plus a typed question and returns a typed decision — a choice, score, or boolean with confidence — positioning it as a drop-in decision layer for classification, routing, scoring, verification, and agent guardrails. The list gives practitioners a fast-scanning field guide to which decision patterns transfer across industries.
60 engagement1 sourcegithubThu, Sep 17, 2026, 02:23 PM
Releases·Tool Release
Simon Willison releases llm-typesafe 0.1a0 plugin bringing TypeSafe AI's Jev model to the LLM CLI
Simon Willison released llm-typesafe 0.1a0, a plugin for his LLM CLI tool that adds support for TypeSafe AI's new Jev model. The plugin supports three answer types — yes/no 'noul' questions, choice questions, and scoring questions — returning structured JSON output. It gives LLM users CLI access to Jev for classification-style tasks.
Simon Willison released llm-typesafe 0.1a0, a plugin for his LLM CLI tool that adds support for TypeSafe AI's new Jev model. The plugin supports three answer types — yes/no 'noul' questions, choice questions, and scoring questions — returning structured JSON output. It gives LLM users CLI access to Jev for classification-style tasks.
0 engagement1 sourcerssTue, Sep 22, 2026, 03:54 PM
Products·launched
Unofficial Jev plugin for coding agents: best practices, API reference, pattern library, and evals
A plugin that lets LLM coding agents (any harness) work with Jev more effectively by providing best practices, anti-patterns, an API reference, and a catalog of 150+ community projects grouped by design pattern with code sketches. It uses progressive disclosure so agents load only the context needed for the current task. The author also ran repeated evals comparing the plugin against no plugin and TypeSafe's official plugin.
A plugin that lets LLM coding agents (any harness) work with Jev more effectively by providing best practices, anti-patterns, an API reference, and a catalog of 150+ community projects grouped by design pattern with code sketches. It uses progressive disclosure so agents load only the context needed for the current task. The author also ran repeated evals comparing the plugin against no plugin and TypeSafe's official plugin.
4 engagement1 sourceredditSun, Sep 27, 2026, 03:55 AM
Developments·Paper
Jev thinks "I don't know'', but doesn't say it: Introducing Sys1Cal-v1 Dataset for Probability Calibration
Two independent arXiv papers examine Jev, a commercial System One model from TypeSafe AI that returns structured decisions with calibrated probabilities instead of text. One benchmarks jev-1.13.0 zero-shot across 37 datasets for under $10, finding it beats Qwen3.8-27B and Gemma-4-E4B on most datasets. The other introduces Sys1Cal-v1, showing Jev's Choice answers hide an implicit third 'I don't know' truth value whose recovery raises median soft accuracy from 0.771 to 0.978.
Two independent arXiv papers examine Jev, a commercial System One model from TypeSafe AI that returns structured decisions with calibrated probabilities instead of text. One benchmarks jev-1.13.0 zero-shot across 37 datasets for under $10, finding it beats Qwen3.8-27B and Gemma-4-E4B on most datasets. The other introduces Sys1Cal-v1, showing Jev's Choice answers hide an implicit third 'I don't know' truth value whose recovery raises median soft accuracy from 0.771 to 0.978.
Updated Tue, Sep 29, 2026, 02:15 PM — Follow-up benchmark paper scores Jev zero-shot across 37 datasets, beating Qwen3.8-27B on 27 of 37 and Gemma-4-E4B on al
0 engagement2 sourcesarxivMon, Sep 28, 2026, 02:59 PM