JEVLAB / FIELD NOTES
Fresh from the lab
A dated reading list of newly added resources and JEVLAB NEWS. Come back for what changed, or follow the RSS feed in your own reader.
Remember this visit on this browser to highlight new additions next time. No sign-in or cross-device tracking.
Dates show when an item became available in JEVLAB, in UTC, not when its original source was published. Translation-only edits do not count as new resources.
36
Instant variable generation
This demonstration showcases Jev's functionality for rapid variable generation within development workflows. The creator highlights how the tool automates data handling, though users should evaluate its performance and integration capabilities within their own specific coding environments.
x.comLeague of Legends Win Probabilities
This project demonstrates a real-time overlay for League of Legends that utilizes Jev to analyze active game states. The creator claims the tool provides live win-probability predictions based on current match data.
x.comjev-oas-sentinel
This tool identifies breaking API changes by combining deterministic checks with TypeSafe JEV System One semantic analysis. It aims to detect inconsistencies often hidden within OpenAPI documentation prose to ensure better API reliability.
github.compi-jev-context
This project demonstrates a reversible context pruning technique for Pi using TypeSafe Jev. It aims to maintain relevant session history while managing memory usage without permanently discarding data.
github.comSimple Jev
Simple Jev is an open-source library designed to convert Hugging Face models into Jev-style endpoints. This tool aims to improve accessibility by simplifying the integration process for developers looking to deploy machine learning models within the Jev ecosystem.
x.comsmart-switch
This macOS utility reimagines window switching by integrating artificial intelligence to predict user intent. The project demonstrates a novel approach to desktop navigation, though specific performance metrics regarding its predictive accuracy remain unverified by independent testing.
github.comJev Gamecast
Jev Gamecast is a React-based application designed for sports enthusiasts. It demonstrates a replay-first architecture that allows users to query live sports data using Jev-typed questions to retrieve specific game insights and historical match information.
github.comCanny
Canny provides a mechanism to prevent AI coding agents from prematurely declaring tasks complete. It utilizes deterministic hooks and an append-only ledger to ensure evidence-based verification of work, operating with zero runtime dependencies.
github.comjev2048
This project demonstrates an implementation of the 2048 puzzle game integrated with Jev. It serves as a practical example of how the TypeSafe AI framework can be applied to automate game logic and decision-making processes.
github.comjev-pii-checker
This command-line tool utilizes TypeSafe Jev to identify personally identifiable information within text. It demonstrates how to detect data presence, assess sensitivity levels, and extract specific character spans for privacy auditing.
github.comJev plays Super Mario Bros.
This demonstration shows Jev interpreting real-time game state data to navigate and play Super Mario Bros. The creator presents this as an example of autonomous decision-making within a classic platformer environment.
x.comJev Piano Composition
This project demonstrates Jev generating piano music by selecting notes and chords based on specific input sequences. It serves as a creative experiment in automated musical composition using Jev architecture.
x.comNanoJev
NanoJev is a compact implementation of the Jev architecture designed to produce full probability distributions within a single forward pass. This tool demonstrates an approach to optimizing model output efficiency for specific computational tasks.
x.comPredicting the next shell command
Jev is a terminal utility designed to forecast upcoming shell commands. It functions by analyzing a user's historical command data to suggest likely inputs, aiming to streamline repetitive command-line workflows.
x.comjev-little-airways
This repository provides a capability study demonstrating the application of Jev, TypeSafe's System One decision model. It serves as an experimental resource to illustrate how the model functions within a specific decision-making framework.
github.comsift
This Chrome extension utilizes TypeSafe Jev to categorize X posts into labels like Substance, Humor, or AI-written. It allows users to filter their feed by hiding content types they prefer to avoid.
github.comjev-skillful
This tool provides a per-prompt capability router for coding agents. It resolves installed skills, MCP servers, and commands using TypeSafe Jev while attempting to measure the effectiveness of these injections on agent performance.
github.comTriage across 1,500 emails
This experiment showcases an automated system designed to classify and manage large volumes of incoming messages. It demonstrates the potential for streamlining inbox organization by processing 1,500 emails to test classification effectiveness.
x.comA Tetris agent
This project showcases an autonomous agent designed to play Tetris. It demonstrates rapid decision-making capabilities, highlighting the potential for real-time game automation through algorithmic play.
x.comtypesafe-sdk-php
This unofficial PHP SDK provides an interface for the TypeSafe AI System One API. The creator claims it maintains functional parity with official JavaScript and Python SDKs, though it is not affiliated with TypeSafe AI.
github.comjev-go
This repository provides an unofficial Go SDK designed to facilitate integration with the TypeSafe AI Jev API. It serves as a developer tool for implementing Jev-based functionality within Go applications.
github.comjev-php-sdk
This SDK enables PHP developers to integrate Jev into their applications. It facilitates sending text and typed queries while retrieving structured answers with confidence scores, supporting PSR-18 clients and various Laravel versions.
github.comORIGIN-CIVILIZATION
This project demonstrates an AI-driven civilization simulation where TypeSafe Jev manages decision-making through typed, probabilistic, and auditable processes. It integrates with various LLMs via OpenAI-compatible APIs to handle high-level planning within the game environment.
github.comjev-browser
This project provides a Jev-powered browser MCP designed for LLM agents. It demonstrates a method for browser interaction that aims to bypass LLM token usage for decision-making, with the creator claiming response times around 300ms.
github.comJev playing Slay the Spire 2
This demonstration showcases the Jev model navigating Slay the Spire 2. The creator claims the model achieves rapid decision-making and high-speed gameplay while processing minimal visual input from the game environment.
x.comsystem-one-adapter-rust
This project provides a Rust implementation of the system-one-adapter, designed to facilitate LLM-backed evaluations. It serves as a technical port for developers looking to integrate system_one evaluation methodologies within a memory-safe Rust environment.
github.comjev-lab
This repository provides TypeScript-based experiments and evaluations for the Jev model. It serves as a technical resource for developers to analyze model performance and latency through structured benchmarks and testing frameworks.
github.comjev-shell-history
This tool integrates Jev-powered ranking into zsh shell history. It demonstrates how to provide Fish-style autosuggestions by leveraging TypeSafe AI to prioritize command history based on user context and relevance.
github.comcomputer_use
This project demonstrates an implementation of MacOS automation controlled by Jev. It serves as an experimental setup for using TypeSafe System One as a decision-making engine to navigate and interact with desktop computer environments.
github.comjevball
jevball is a 3D football simulation where each player operates as an independent Jev decision-making agent. The project demonstrates multi-agent coordination in a sports context, allowing users to observe autonomous gameplay or manually control a specific player.
github.comWinnow
Winnow is an open-source Chrome extension that analyzes articles and YouTube videos to provide reading recommendations. It uses Jev technology to categorize content as read, skim, save, or skip based on user-defined goals.
github.comactiongate-jev
This project provides runtime authorization and guardrails for AI-agent tool calls. It utilizes deterministic policies and TypeSafe Jev integration via OpenRouter to manage agent interactions securely.
github.comEgo Lite Shopping Agent
This demonstration showcases an agent utilizing Ego Lite and Jev to perform rapid product evaluations. The creator claims the system completes twenty distinct shopping assessments in under four seconds.
x.comjevmod
jevmod provides a framework for automated content moderation by applying category-specific probability thresholds. This tool allows developers to implement customizable filtering logic to manage and classify digital content effectively.
github.comjevibe-check
This tool provides real-time tone analysis for Bluesky posts and drafts. It demonstrates an integration with the TypeSafe Jev API to automatically label the emotional sentiment of text content before publication.
github.comVercel AI SDK Provider
This resource introduces the typesafe-ai provider for the Vercel AI SDK. It demonstrates how to utilize the experimental_evaluate function while incorporating jev-latest as an evaluation model for testing AI outputs within a development environment.
ai-sdk.dev
