Jev's Garage
A garage full of tiny experiments for building critical systems with TypeSafe AI's Jev/System One, including demos, SDK integration, and live contract tests.
OPEN SOURCE · CURATED REPOSITORIES
Curated Jev SDKs, tools, demos, integrations, benchmarks, and research projects. Repository metrics come from GitHub; titles, summaries, and categories are editorial.
Ranked by GitHub Stars
Ranked by recent Star growth and update activity
A garage full of tiny experiments for building critical systems with TypeSafe AI's Jev/System One, including demos, SDK integration, and live contract tests.
A learning project that implements a TypeSafe Jev/System One-style decision endpoint on a Qwen3-0.6B backbone with LoRA, plus an OpenAI-compatible chat/generation server and MiniJev-compatible evaluation. Not production software.
Minesweeper experiments comparing TypeSafe Jev (System One), LLMs, hybrid policies, and deterministic solvers, with documented methods and recorded results.
Six recorded chess experiments that call TypeSafe Jev through OpenRouter: independent play, a tactical filter, Stockfish assistance, and a Jev review loop, with full traces, replays, and cost records.
A 2048 web lab where every move is driven by TypeSafe AI's Jev decision model, showing probability, confidence, latency, and cost, with no heuristic fallback.
Two System One models fight a real Doom deathmatch using the same state, typed questions, and shield; Jev is TypeSafe's hosted jev-1.13.
Semantic pub/sub experiment using TypeSafe AI's Jev System One model. Routes messages by meaning—one batched request evaluates all subscribers in parallel. Includes measured stability, batch degradation, and wording sensitivity tests.
An empirical experiment and API research repository for TypeSafe's Jev / System One model, running controlled tests to evaluate model behavior and saving raw evidence.
Connects TypeSafe Jev to Pokémon Showdown, recording 40 battles and 940 API decisions with reproducible results, datasets, and annotated replays. Independent research, not an official benchmark.
An open lab that puts TypeSafe's Jev (System One) in front of Claude Code, with measured silent bugs, a patch, and a hard fallback with alerts.
An experimental project using TypeSafe JEV to forecast short-term Hong Kong stock direction, with data fetching, state construction, backtesting, and report generation.
An intrusion detection prototype using TypeSafe's Jev System One model, benchmarked on NSL-KDD against LLM and Random Forest baselines.