Skip to content

Jev directory

Jevscript

A programming language where judgment is a runtime primitive: programs ask Jev narrow typed questions (noul, choice, score) as they run, and the Hocket engine coalesces judgments so two asks over one state become a single API request.
Stars
1
Language
JavaScript
Reading time
2 min
Published by
Community
Author
amberwhitehead
Last verified
Oct 2026

Highlights

  • JS-ish surface with three question forms (noul, choice, score) returning typed answers that carry probability and confidence.
  • Hocket coalesces lazy judgments: everything built over the same state flushes as one request, and answers are memoized on question identity and state hash.
  • The README reports about 2.7x input-token savings at a ~230-character state (measured at M0) and cites roughly 12x on document-dominated workloads.
  • Status M0 (2026-09-16): the spike confirmed the answer contract and recorded the batching baseline; parser and interpreter are planned for M1.
  • The deployment model keeps the API key server-side through a zero-dependency proxy pattern inherited from the sibling typesafe_test car-driver project.

Watch out

Copy reviewed against the repo README on 2026-10-15. The project is at milestone M0: the language, parser and interpreter are still milestones ahead, and GitHub metadata lists no license, so the code is all rights reserved by default.

Questions about this entry

Short answers drawn from the linked project or page.

From 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.

Follow @typesafeai