Launches on September 17, 2026

Kotlin Multiplatform SDK for the TypeSafe / Jev System One API — typed questions with calibrated probabilities instead of generated text.

Six small apps and a workbench that show what TypeSafe's Jev (System One) model can do

Lint a Markdown knowledge base (Obsidian vault or LLM wiki) for contradictions, stale claims, unresolved markers and missing pages, using TypeSafe Jev

A visible 2048 experiment powered by TypeSafe

a multi-chain EVM token screener built on Jev by @TypeSafe AI

Go 1.27 client for TypeSafe AI's Jev (System One) API: typed questions, typed answers

Real-vaxt taksi dispetçerlik simulyasiyası - TypeSafe JEV (System One) modeli ilə

Public repo containing generated synthetic data to test jev from typesafe

A forward-only experiment: Jev (TypeSafe System One) making one-second trading judgments on bitbank, logged for calibration analysis.

Trading project using JEV by TypeSafe AI

One AI trade decision every Monad block. Jev on Kuru MON-USDC.

TypeSafe Jev demos: Play inbox, Czajka guard, agent-card router, seed comparator, RL data triage

One MCP server for TypeSafe's Jev: typed, calibrated decisions instead of generated text

Checking JEV's Contradiction detection (Model by TypeSafe.AI)

TypeSafe Jev as a bounded judgement layer for computer use: verify, pick, classify over Cua Driver and phone-harness

Unofficial MCP for Jev

An open-source decision layer connecting coding agents with TypeSafe Jev.

Reproducible text-first browser automation experiment with TypeSafe Jev and Playwright.

Jev-compatible decision engine over caller-provided models

Local, single-user job-posting evaluator: typed semantic signals plus a deterministic scorer produce explainable 0-100 fit scores, with every claim traced to a line of the posting.

Conversational agent workspace with guided onboarding, voice, evaluations and monitoring.

棋聚 Kee Club — Jungle and Aeroplane Chess with friends, AI opponents, and offline same-device play.

Portable benchmark for LLM tier routing, factual-claim verification, and safe interim messages

Provider-agnostic LLM client, adapters (Anthropic/OpenAI/Google), and tool-calling agent harness for Go

Tkinter chat and live visual narration with LM Studio, TypeSafe Jev, and Chatterbox TTS

Local open-source probabilistic decision inference for open-weight LLMs using direct candidate scoring.

Neuro inspired architecture for improving OCR in low light or low resolution settings.

A TypeSafe Jev-powered Magic 8 Ball

Experimental Facebook Messenger message-request triage using AI classification and guarded reply drafts. Uses browser automation, not Meta API; does not auto-accept requests or reply to unaccepted threads.

Booking-to-SPA conversion intelligence for property developers (YEI 3.0, Chin Hin Group)

NightDesk: session-aware buy screen for tokenized stocks on BSC (BNB Hack). Truth card + TypeSafe execution guard.

opencode plugin: send the model only the MCP tools each request needs, picked by TypeSafe's Jev

TypeSafe-powered document classification for Paperless-ngx: existing tags, document types, and reviewable decisions.

A living index of Physical AI research, organized by domain and lab.

A bounded TypeSafe Jev decision tool for Pi

Jev-powered semantic process conditions and skill suggestions for Pi

Pi extension: classify every tool call with TypeSafe's Jev (allow / ask / deny) so the main model never spends tokens judging its own tool calls

Pi extension: TypeSafe Jev routes turns, pre-fetches context, trims tool results, catches loops and guards tool calls

Open-source generation-ship command simulation with Astra and Jev.

RAG Gatekeeper — TypeSafe chunk score/filter before LLM

Agent Skills-compatible skill: govern a Raindrop.io library's collection structure via raindrop-mcp — audit, confirmed restructuring & misplaced-bookmark relocation. Works with Claude Code, Codex CLI, Gemini CLI, GitHub Copilot, Cursor & WorkBuddy.

Jev-powered execution state for Pi coding agents. Track changes, checks, and blockers outside the main LLM, with evidence-backed updates and replay. Inspired by SKILL.state.

Evidence-aware local RAG and HandoffProof: controlled causal testing for operational handovers, with optional TypeSafe/Jev evidence governance.

Terminal LAN monitor: local rules as senses, TypeSafe Jev as risk brain. Speed/accuracy experiment.

A local-first skill library and visual skill tree, with Jev routing and MCP access.

Human-in-the-loop AI agent workflow for supervised sales automation, approvals and agent decision systems.

Turn any open model into a fast, typed, calibrated decision function (parallel scoring, no generation)

An LLM router that uses Jev (TypeSafe System One) to dispatch each request to the cheapest model and lowest reasoning effort that will do the job. OpenAI-compatible proxy, Claude Code and Codex wrappers, TypeScript and C#.

Playful ticket triage that shows how TypeSafe Jev's typed, probability-backed decisions compare with prompting an LLM. Cost of Delay ranking, certainty-gated routing, bring-your-own-key LLM race.

FastAPI service that reranks support tickets by urgency using TypeSafe (Jev / System One)