A source-backed index of Jev projects, demos, integrations, articles and benchmarks. https://jevland.mnimiy.com/ 2026-09-22T00:00:00Z https://jevland.mnimiy.com/projects/pluvio-twenty-jev-projects-cn/ 2026-09-21T00:00:00Z 2026-09-21T00:00:00Z 雪踏乌云 (Pluvio9yte)

A Chinese list of twenty projects for people who just got an API key: jev-ultrafast, fast-jev-compaction, json-render, typesafe-mcp, jev-mcp, SemDecide, the Codex router, winnow, jev-review, blink, Agent Desktop, typesafe-mario, jev-drone, OneVOneJev, jev-trader, prism-liquidity-agent, neo4jev, jev-curate, Canny and killmyidea.

<p>A Chinese list of twenty projects for people who just got an API key: jev-ultrafast, fast-jev-compaction, json-render, typesafe-mcp, jev-mcp, SemDecide, the Codex router, winnow, jev-review, blink, Agent Desktop, typesafe-mario, jev-drone, OneVOneJev, jev-trader, prism-liquidity-agent, neo4jev, jev-curate, Canny and killmyidea.</p><p><strong>commentary:</strong> Roundup with twenty repository links.</p><p>X post only.</p><p><a href="https://x.com/Pluvio9yte/status/2101831273224311035">Read original post on X</a></p> https://jevland.mnimiy.com/projects/viv-jev-as-judge-rl-verifiers/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Viv (Vtrivedy10)

Many RL tasks need a judge to verify pieces of outputs or trajectories, and at scale that is the bottleneck; a fast, cheap, calibrated judge could remove it. Draws on earlier work with Harvey's LAB benchmark, where harness engineering plus open models cut judging costs by orders of magnitude.

<p>Many RL tasks need a judge to verify pieces of outputs or trajectories, and at scale that is the bottleneck; a fast, cheap, calibrated judge could remove it. Draws on earlier work with Harvey&#39;s LAB benchmark, where harness engineering plus open models cut judging costs by orders of magnitude.</p><p><strong>commentary:</strong> Commentary on LangChain&#39;s judge post; calibration caveat from the author.</p><p>X post only.</p><p><a href="https://x.com/Vtrivedy10/status/2101688022513160700">Read original post on X</a></p> https://jevland.mnimiy.com/projects/jev-in-front-of-grok-build/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Dan (Daniel_Farinax)

An X Article on putting Jev as a gate in front of Grok Build and measuring the cost difference on identical tasks.

<p>An X Article on putting Jev as a gate in front of Grok Build and measuring the cost difference on identical tasks.</p><p><strong>author demo:</strong> 22–40% cheaper per the author&#39;s runs.</p><p>X post only.</p><p><a href="https://x.com/Daniel_Farinax/status/2101749959980728575">Read article on X</a></p> https://jevland.mnimiy.com/projects/keno-captcha-arbitrage/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z keno

A provocation with a video: if a marketplace pays $0.01 per CAPTCHA and Jev solves 100 for $0.0068, the margin is 99%; the author threatens to open-source it if the loophole is not fixed.

<p>A provocation with a video: if a marketplace pays $0.01 per CAPTCHA and Jev solves 100 for $0.0068, the margin is 99%; the author threatens to open-source it if the loophole is not fixed.</p><p><strong>author demo:</strong> Author&#39;s numbers; no code released. Abuse-adjacent — medium risk.</p><p>X post only.</p><p><a href="https://x.com/kenonews/status/2101656436136661163">Read original post on X</a></p> https://jevland.mnimiy.com/projects/burhan-video-clipping/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Burhan Usman

Transcript in, topic in, clips out: Jev picks the segments of a 90-minute video for any topic in under two seconds.

<p>Transcript in, topic in, clips out: Jev picks the segments of a 90-minute video for any topic in under two seconds.</p><p><strong>author demo:</strong> Under 2 seconds and about 2 cents per run, per the author.</p><p>X post only.</p><p><a href="https://x.com/BurhanUsman/status/2101641842441732297">Read original post on X</a></p> https://jevland.mnimiy.com/projects/y0usaf-pi-jev-tool-gate/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z y0usaf

Jev as a decision layer for the Pi coding agent: a measured tool-call gate plus a jev_ask tool for typed, calibrated answers; distinct from madeye/pi-jev.

<p>Jev as a decision layer for the Pi coding agent: a measured tool-call gate plus a jev_ask tool for typed, calibrated answers; distinct from madeye/pi-jev.</p><p><strong>source-backed:</strong> Repository; no figures.</p><p><a href="https://github.com/y0usaf/pi-jev">Open repository</a></p> https://jevland.mnimiy.com/projects/open-alternative-jev/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z ikermoel

A System One style layer that returns typed, calibrated decisions from any open-weights LLM in a single forward pass (Hugging Face and vLLM), with benchmarks against Jev in the README.

<p>A System One style layer that returns typed, calibrated decisions from any open-weights LLM in a single forward pass (Hugging Face and vLLM), with benchmarks against Jev in the README.</p><p><strong>source-backed:</strong> Benchmarks published in the repository.</p><p><a href="https://github.com/ikermoel/open-alternative-jev">Open repository</a></p> https://jevland.mnimiy.com/projects/reticle-runtime-perception/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Reticle

Agents can write code but struggle to understand what they built at runtime; Reticle gives web and desktop apps a Jev-style runtime perception layer so agents know which interface they are looking at.

<p>Agents can write code but struggle to understand what they built at runtime; Reticle gives web and desktop apps a Jev-style runtime perception layer so agents know which interface they are looking at.</p><p><strong>source-backed:</strong> Repository; the team also published the Jev-vs-LLM-judge cost comparison on its blog.</p><p><a href="https://github.com/reticlehq/reticle">Open repository</a></p> https://jevland.mnimiy.com/projects/jev-browser-llm-plans-jev-decides/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Ying-Kai Liao

Browser automation split in two: an LLM plans the task, Jev makes the concrete in-page choices; shipped as a library, a CLI and an MCP server.

<p>Browser automation split in two: an LLM plans the task, Jev makes the concrete in-page choices; shipped as a library, a CLI and an MCP server.</p><p><strong>source-backed:</strong> Repository; no figures.</p><p><a href="https://github.com/Ying-Kai-Liao/jev-browser">Open repository</a></p> https://jevland.mnimiy.com/projects/canny-evidence-before-done/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z qkal

Deterministic hooks decide, Jev advises: an append-only ledger with zero runtime dependencies that blocks an agent's 'done' until there is evidence.

<p>Deterministic hooks decide, Jev advises: an append-only ledger with zero runtime dependencies that blocks an agent&#39;s &#39;done&#39; until there is evidence.</p><p><strong>source-backed:</strong> Repository; no figures.</p><p><a href="https://github.com/qkal/Canny">Open repository</a></p> https://jevland.mnimiy.com/projects/jev-curate-dataset-sifter/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Akash Priyadarshi

A high-throughput sifter for synthetic and pretraining datasets: stream Parquet and JSONL, filter and score every row with Jev.

<p>A high-throughput sifter for synthetic and pretraining datasets: stream Parquet and JSONL, filter and score every row with Jev.</p><p><strong>source-backed:</strong> 1,500+ rows per second per the README.</p><p><a href="https://github.com/AkashPriyadarshii/jev-curate">Open repository</a></p> https://jevland.mnimiy.com/projects/prism-liquidity-agent/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z irfndi

An autonomous liquidity-provider agent with auto-rebalancing and backtested strategies (Meteora DLMM first), with Jev in the decision path.

<p>An autonomous liquidity-provider agent with auto-rebalancing and backtested strategies (Meteora DLMM first), with Jev in the decision path.</p><p><strong>source-backed:</strong> Repository with backtests. Finance — high-risk category.</p><p><a href="https://github.com/irfndi/prism-liquidity-agent">Open repository</a></p> https://jevland.mnimiy.com/projects/one-v-one-jev-arena/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z emrickgarrett

A browser 1v1 quickscope arena built with Three.js where Jev plays the opponent's decisions in real time.

<p>A browser 1v1 quickscope arena built with Three.js where Jev plays the opponent&#39;s decisions in real time.</p><p><strong>source-backed:</strong> Playable repository.</p><p><a href="https://github.com/emrickgarrett/OneVOneJev">Open repository</a></p> https://jevland.mnimiy.com/projects/agent-desktop-rust-accessibility/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z lahfir

A Rust runtime that lets any agent operate desktop apps through their real UI structure (OS accessibility trees) with stable element refs, from the author of the hook-based routing demo.

<p>A Rust runtime that lets any agent operate desktop apps through their real UI structure (OS accessibility trees) with stable element refs, from the author of the hook-based routing demo.</p><p><strong>source-backed:</strong> Repository; no figures.</p><p><a href="https://github.com/lahfir/agent-desktop">Open repository</a></p> https://jevland.mnimiy.com/projects/itsmostafa-typesafe-mcp/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z itsmostafa

An MCP connector that gives any MCP-capable agent direct Choice / Score / Noul calls to Jev; the Chinese roundups call it the first thing to install after getting an API key.

<p>An MCP connector that gives any MCP-capable agent direct Choice / Score / Noul calls to Jev; the Chinese roundups call it the first thing to install after getting an API key.</p><p><strong>source-backed:</strong> MCP connector; no figures.</p><p><a href="https://github.com/itsmostafa/typesafe-mcp">Open repository</a></p> https://jevland.mnimiy.com/projects/semdecide-unix-pipelines/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z sharziki

Jev as a command-line filter: classify, score and filter lines in a shell pipeline, for crawlers, CI gates and data pipelines.

<p>Jev as a command-line filter: classify, score and filter lines in a shell pipeline, for crawlers, CI gates and data pipelines.</p><p><strong>source-backed:</strong> CLI; no figures.</p><p><a href="https://github.com/sharziki/semdecide">Open repository</a></p> https://jevland.mnimiy.com/projects/winnow-context-sieve/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Ghaleb Dweikat

Every tool result is judged by a System One model before it enters Claude Code's context, so noise never gets a chance to bloat the window.

<p>Every tool result is judged by a System One model before it enters Claude Code&#39;s context, so noise never gets a chance to bloat the window.</p><p><strong>source-backed:</strong> Repository; found via a Chinese roundup.</p><p><a href="https://github.com/GhalebDweikat/winnow">Open repository</a></p> https://jevland.mnimiy.com/projects/hermes-jev-skills/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Steve Darlow (kerpopule)

Jev-powered model routing, memory, context compaction, skill selection and computer/browser use for Hermes agents (also Claude Code and Codex), with a routing dashboard and shadow mode that shows what Jev routing would have saved before you switch it on.

<p>Jev-powered model routing, memory, context compaction, skill selection and computer/browser use for Hermes agents (also Claude Code and Codex), with a routing dashboard and shadow mode that shows what Jev routing would have saved before you switch it on.</p><p><strong>source-backed:</strong> ~0.4 s per decision, fractions of a cent; author reports an 81.8% cache hit rate on his fleet and that above ~32k context a downgrade never pays because of cache rebuild.</p><p><a href="https://github.com/kerpopule/hermes-jev-skills">Open repository</a></p><p><a href="https://x.com/StevenDarlow/status/2101526148228227519">Read original post on X</a></p> https://jevland.mnimiy.com/projects/docjev-document-classification/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Jerry Liu (LlamaIndex)

An OSS library from the LlamaIndex founder: give a document plus natural-language category rules and Jev predicts the category (classify) or the boundaries between sub-documents (split). OCR backends: liteparse (fast, free) or LlamaParse for complex layouts.

<p>An OSS library from the LlamaIndex founder: give a document plus natural-language category rules and Jev predicts the category (classify) or the boundaries between sub-documents (split). OCR backends: liteparse (fast, free) or LlamaParse for complex layouts.</p><p><strong>source-backed:</strong> 6x faster than gpt-5.6-luna with equivalent accuracy, inclusive of liteparse time; interactive report in docs/report.</p><p><a href="https://github.com/jerryjliu/docjev">Open repository</a></p><p><a href="https://x.com/jerryjliu0/status/2101738281046294552">Read original post on X</a></p> https://jevland.mnimiy.com/projects/openrouter-playground-jev/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Alex Atallah (OpenRouter)

OpenRouter's founder pointing to the model page where Jev 1.13 can be tried directly in the playground.

<p>OpenRouter&#39;s founder pointing to the model page where Jev 1.13 can be tried directly in the playground.</p><p><strong>official:</strong> Platform page.</p><p><a href="https://openrouter.ai/typesafe/jev-1.13">Open service</a></p><p><a href="https://x.com/alexatallah/status/2101500888065974307">Read original post on X</a></p> https://jevland.mnimiy.com/projects/greptile-realtime-bug-checker/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Daksh Gupta (Greptile)

A simple real-time bug checker built with Jev for people who still code by hand; it also underlines inaccurate English sentences.

<p>A simple real-time bug checker built with Jev for people who still code by hand; it also underlines inaccurate English sentences.</p><p><strong>author demo:</strong> Live page; no figures.</p><p><a href="https://greptile.com/realtime">Open live demo</a></p><p><a href="https://x.com/dakshgup/status/2101678769022841297">Read original post on X</a></p> https://jevland.mnimiy.com/projects/langchain-webinar-building-a-harness-with-jev/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z LangChain (Harrison Chase)

LangChain's webinar on using Jev to build a better agent harness, with Sydney Runkle, Hunt Lovell and Allie.

<p>LangChain&#39;s webinar on using Jev to build a better agent harness, with Sydney Runkle, Hunt Lovell and Allie.</p><p><strong>official:</strong> Event page; registration on events.langchain.com.</p><p><a href="https://events.langchain.com/webinar/building-a-harness-with-jev/">Read official post</a></p><p><a href="https://x.com/hwchase17/status/2101670779926499683">Read original post on X</a></p> https://jevland.mnimiy.com/projects/agent-arena-degen-village-bouncer/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z zostaff

A 'bouncer called Jev' added to an agent trading arena that blocks bad trades before the AI even sees them.

<p>A &#39;bouncer called Jev&#39; added to an agent trading arena that blocks bad trades before the AI even sees them.</p><p><strong>source-backed:</strong> Author&#39;s tally: 23 bad trades blocked in one night. Finance — high-risk category.</p><p><a href="https://github.com/zostaff/agent-arena">Open repository</a></p><p><a href="https://x.com/zostaff/status/2101693557664862537">Read original post on X</a></p> https://jevland.mnimiy.com/projects/nerve-onchain-guard/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z h100envy

A trading agent guard where Jev vets trades before the LLM commits; the author says it stopped 3 ETH going into dead pools in the first five days.

<p>A trading agent guard where Jev vets trades before the LLM commits; the author says it stopped 3 ETH going into dead pools in the first five days.</p><p><strong>source-backed:</strong> Author&#39;s own tally (&#39;saved, not made&#39;). Finance — high-risk category.</p><p><a href="https://github.com/h100envy/nerve">Open repository</a></p><p><a href="https://x.com/h100envy/status/2101688318950068263">Read original post on X</a></p> https://jevland.mnimiy.com/projects/semgate-go-middleware/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z m-mizutani

Go HTTP middleware that lets Jev judge and route requests, plus a CTF-style example site inviting people to try injection attacks against it.

<p>Go HTTP middleware that lets Jev judge and route requests, plus a CTF-style example site inviting people to try injection attacks against it.</p><p><strong>source-backed:</strong> Middleware and a live CTF site (Cloud Run).</p><p><a href="https://semgate-example-90572511273.asia-northeast1.run.app/">Open live demo</a></p><p><a href="https://x.com/m_mizutani/status/2101500133846430086">Read original post on X</a></p> https://jevland.mnimiy.com/projects/hunk-triage-diff-classification/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z morinokami

An extension for bentlegen's hunk that classifies the files in a diff (core, tests, docs…) with Jev and reorders the review so the important changes come first.

<p>An extension for bentlegen&#39;s hunk that classifies the files in a diff (core, tests, docs…) with Jev and reorders the review so the important changes come first.</p><p><strong>source-backed:</strong> Classification-only extension; no figures.</p><p><a href="https://github.com/morinokami/hunk-triage">Open repository</a></p><p><a href="https://x.com/onDemocracy/status/2101654957195931740">Read original post on X</a></p> https://jevland.mnimiy.com/projects/jev-lexer-syntax-highlighter/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z mizchi

Split like gpu-lexer, classify every token with Jev, render Shiki-compatible tokens to HTML and ANSI — a language-agnostic highlighter, experimental only.

<p>Split like gpu-lexer, classify every token with Jev, render Shiki-compatible tokens to HTML and ANSI — a language-agnostic highlighter, experimental only.</p><p><strong>source-backed:</strong> About 91.75% token accuracy and roughly $0.37 for a 1,000-line file, per the author.</p><p><a href="https://github.com/mizchi/jev-lexer">Open repository</a></p><p><a href="https://x.com/mizchi/status/2101623687778828467">Read original post on X</a></p> https://jevland.mnimiy.com/projects/tab-bouncer/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z MANISH007700

Tell the extension what you are working on; Jev checks every open tab at the door in a single call and shows the freeloaders out.

<p>Tell the extension what you are working on; Jev checks every open tab at the door in a single call and shows the freeloaders out.</p><p><strong>source-backed:</strong> One Jev call per sweep; no figures.</p><p><a href="https://github.com/MANISH007700/tab-bouncer">Open repository</a></p><p><a href="https://x.com/lucifer_x007/status/2101574162817093655">Read original post on X</a></p> https://jevland.mnimiy.com/projects/rmalde-minecraft-agent/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z rmalde

A Minecraft agent that plans with Astra and lets Jev control step-by-step actions, with native recording, tested routes and run verification; the Jev decision rules live in models.mjs.

<p>A Minecraft agent that plans with Astra and lets Jev control step-by-step actions, with native recording, tested routes and run verification; the Jev decision rules live in models.mjs.</p><p><strong>source-backed:</strong> Repository with tested routes; a Japanese reader notes the decision criteria are written out in natural language.</p><p><a href="https://github.com/rmalde/minecraft-agent">Open repository</a></p><p><a href="https://x.com/cled0328/status/2101579523573928395">Read original post on X</a></p> https://jevland.mnimiy.com/projects/mapika-decider-2b/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z angework

A Jev-style 2B decider tested on a consumer RTX 5070 Ti under Windows 11.

<p>A Jev-style 2B decider tested on a consumer RTX 5070 Ti under Windows 11.</p><p><strong>author demo:</strong> After optimisation: 10.50 ms average, 95 decisions per second, 3.53 GiB VRAM on short two-question prompts ×100.</p><p><a href="https://huggingface.co/Mapika/decider-2b">Open service</a></p><p><a href="https://x.com/angeworkCoLtd/status/2101568907366785297">Read original post on X</a></p> https://jevland.mnimiy.com/projects/openthai-systemone/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z iApp Technology (kobkrit)

Five days after launch: a Thai + English System One model on Qwen3.5-0.8B with a 256-way slot head, Apache-2.0, sized to run on a laptop.

<p>Five days after launch: a Thai + English System One model on Qwen3.5-0.8B with a 256-way slot head, Apache-2.0, sized to run on a laptop.</p><p><strong>source-backed:</strong> 0.8B parameters, Apache-2.0; figures in the repository.</p><p><a href="https://github.com/iapp-technology/openthai-systemone">Open repository</a></p><p><a href="https://x.com/kobkrit/status/2101699356810453448">Read original post on X</a></p> https://jevland.mnimiy.com/projects/open-jev-typed-decision-engine/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z intikhab49

An open reproduction that answers noul/choice/score in one non-autoregressive pass with calibrated confidence, at 150M parameters — a non-LLM route to the same interface.

<p>An open reproduction that answers noul/choice/score in one non-autoregressive pass with calibrated confidence, at 150M parameters — a non-LLM route to the same interface.</p><p><strong>source-backed:</strong> README: 0.697 vs Jev&#39;s 0.727 on its eval, 2.5x better calibrated, 4x faster.</p><p><a href="https://github.com/intikhab49/open-jev-typed-decision-engine">Open repository</a></p><p><a href="https://x.com/sounisi5011/status/2101643782936830369">Read original post on X</a></p> https://jevland.mnimiy.com/projects/laya-mlx-and-coreml/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z mizorewww

Native MLX and Core ML runtimes for Laya typed decisions: `pip install 'laya-mlx[demo]'`, no text generation, no PyTorch, no cloud API. A Rubik's-cube demo runs 100 moves locally.

<p>Native MLX and Core ML runtimes for Laya typed decisions: `pip install &#39;laya-mlx[demo]&#39;`, no text generation, no PyTorch, no cloud API. A Rubik&#39;s-cube demo runs 100 moves locally.</p><p><strong>source-backed:</strong> 7–14 ms short decisions on an M3 Max (MLX) and ~5 ms on Core ML / Neural Engine, per the repositories&#39; benchmarks.</p><p><a href="https://github.com/mizorewww/laya-mlx">Open repository</a></p><p><a href="https://x.com/Delroy715/status/2101530086671352235">Read original post on X</a></p> https://jevland.mnimiy.com/projects/jev-as-a-judge-langchain/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z LangChain (Daniel Shea, Sean)

LangChain tested Jev against LLM judges on accuracy, repeatability, latency and cost to see whether System One models offer a new approach to online agent evaluation; evaluator code in the repository.

<p>LangChain tested Jev against LLM judges on accuracy, repeatability, latency and cost to see whether System One models offer a new approach to online agent evaluation; evaluator code in the repository.</p><p><strong>official:</strong> Repository with the evaluator; the post calls Jev cheaper and more precise than LLM judges for online evals.</p><p><a href="https://github.com/danielgshea/jev-as-a-judge">Open repository</a></p><p><a href="https://x.com/LangChain/status/2101454284927959080">Read original post on X</a></p> https://jevland.mnimiy.com/projects/jev-paper-judge/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z JacobLinCool

Upload a research-paper PDF; text is extracted in the browser, split into roughly 1,000-word passages, and a fixed set of small questions is asked per passage in parallel, then combined into weighted scores.

<p>Upload a research-paper PDF; text is extracted in the browser, split into roughly 1,000-word passages, and a fixed set of small questions is asked per passage in parallel, then combined into weighted scores.</p><p><strong>author demo:</strong> Free demo of the many-small-questions-combined-in-code pattern.</p><p><a href="https://jev-paper-judge.jacob.workers.dev">Open live demo</a></p> https://jevland.mnimiy.com/projects/rubikjev-cube-solver/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Trou (0xtrou)

Scramble a Rubik's cube and watch Jev judge and solve it live, with difficulty tiers, XP and token-metered runs; labelled experimental and fan-made.

<p>Scramble a Rubik&#39;s cube and watch Jev judge and solve it live, with difficulty tiers, XP and token-metered runs; labelled experimental and fan-made.</p><p><strong>author demo:</strong> Demo.</p><p><a href="https://rubikjev.solo.engineer">Open live demo</a></p> https://jevland.mnimiy.com/projects/last-exit-checkpoint-game/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z 0x963D

A text game where you talk your way past a border inspector whose actions are chosen by Jev from your typed replies; dialogue and consequences are authored game code, Jev only decides.

<p>A text game where you talk your way past a border inspector whose actions are chosen by Jev from your typed replies; dialogue and consequences are authored game code, Jev only decides.</p><p><strong>author demo:</strong> Hosted demo; may run out of demo credits.</p><p><a href="https://gate.fade.tools">Open live demo</a></p> https://jevland.mnimiy.com/projects/typesafe-image-diffusion-pixel-art/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Aryan Singh

Turns Jev into a pixel-art generator by asking 256 multiple-choice questions, one per pixel, plus refinement passes.

<p>Turns Jev into a pixel-art generator by asking 256 multiple-choice questions, one per pixel, plus refinement passes.</p><p><strong>author demo:</strong> Playful demo.</p><p><a href="https://github.com/Wizhill05/typesafe-image-diffusion">Open repository</a></p><p><a href="https://x.com/just_aryansingh/status/2100617080395710748">Read original post on X</a></p> https://jevland.mnimiy.com/projects/jev-secret-detection-test/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Teyhouse

A quick credential and secret-detection experiment with Jev; the author reports consistent results across runs.

<p>A quick credential and secret-detection experiment with Jev; the author reports consistent results across runs.</p><p><strong>author demo:</strong> Demo.</p><p><a href="https://github.com/teyhouse/jev-secret-detection">Open repository</a></p><p><a href="https://x.com/Teyhouse/status/2100555273718907064">Read original post on X</a></p> https://jevland.mnimiy.com/projects/jevmeter-live-fact-checker/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Chetaslua

An open-source tool that fact-checks spoken content (debates, investor calls, interviews, podcasts) live with Jev calls; demoed on an interview.

<p>An open-source tool that fact-checks spoken content (debates, investor calls, interviews, podcasts) live with Jev calls; demoed on an interview.</p><p><strong>author demo:</strong> Demo run: 60 Jev calls, 111K tokens, $0.0047.</p><p><a href="https://github.com/ChetasLua/jevmeter">Open repository</a></p><p><a href="https://x.com/chetaslua/status/2100602714204049588">Read original post on X</a></p> https://jevland.mnimiy.com/projects/ground-truth-news-framing/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z D (jagenaujagenau)

A browser extension (and web version) that classifies the news article you are reading by political framing, article type, topic and how loaded the language is, using Jev as a fast probabilistic classifier.

<p>A browser extension (and web version) that classifies the news article you are reading by political framing, article type, topic and how loaded the language is, using Jev as a fast probabilistic classifier.</p><p><strong>author demo:</strong> Open source.</p><p><a href="https://groundtruth.click/">Open service</a></p><p><a href="https://x.com/jagenaujagenau/status/2100622352333574460">Read original post on X</a></p> https://jevland.mnimiy.com/projects/agent-handoff-gate/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Matt (zxdubx)

An experimental protocol where a worker agent sends requirements, code and test results to Jev, which returns an evidence verdict for the team-lead agent before results are accepted.

<p>An experimental protocol where a worker agent sends requirements, code and test results to Jev, which returns an evidence verdict for the team-lead agent before results are accepted.</p><p><strong>author demo:</strong> Experimental; Jev judges evidence, the policy grants permissions.</p><p><a href="https://github.com/zsoXi/agent-handoff-gate">Open repository</a></p><p><a href="https://x.com/zxdubx/status/2100604919120121960">Read original post on X</a></p> https://jevland.mnimiy.com/projects/mcp-jev-packs/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z pedroknigge

A local MCP server that runs Jev "packs" (code audit, PR review, GUI automation) for Cursor, Claude Code and Codex.

<p>A local MCP server that runs Jev &quot;packs&quot; (code audit, PR review, GUI automation) for Cursor, Claude Code and Codex.</p><p><strong>author demo:</strong> About 6,000 files scanned in roughly two minutes with eight workers.</p><p><a href="https://github.com/pedroknigge/mcp_jev">Open repository</a></p> https://jevland.mnimiy.com/projects/flue-jev-cloudflare-demo/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Matthew Phillips

A Cloudflare Worker demo where a Flue agent uses Jev through Cloudflare AI Gateway for intent and urgency routing.

<p>A Cloudflare Worker demo where a Flue agent uses Jev through Cloudflare AI Gateway for intent and urgency routing.</p><p><strong>author demo:</strong> Demo.</p><p><a href="https://github.com/matthewp/flue-jev-demo">Open repository</a></p> https://jevland.mnimiy.com/projects/invalidate-memory-leases/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z chopratejas

A layer that decides when an agent's stored memory should be invalidated, using Jev checks; a playground is hosted.

<p>A layer that decides when an agent&#39;s stored memory should be invalidated, using Jev checks; a playground is hosted.</p><p><strong>author demo:</strong> 89.2% strict accuracy on 157 cases, about $0.00006 and 150 ms per check, zero false invalidations.</p><p><a href="https://invalidate-playground.vercel.app">Open live demo</a></p> https://jevland.mnimiy.com/projects/jevql-postgres-predicates/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Kyle McLaren

Adds `jev()` semantic predicates to plain Postgres through a CLI, HTTP and SDKs, with a SQLite cache and an `--explain` cost preview; a playground is hosted.

<p>Adds `jev()` semantic predicates to plain Postgres through a CLI, HTTP and SDKs, with a SQLite cache and an `--explain` cost preview; a playground is hosted.</p><p><strong>author demo:</strong> Third Postgres-flavoured integration after pg-jev and pg_typesafe.</p><p><a href="https://jevql.fly.dev/">Open live demo</a></p> https://jevland.mnimiy.com/projects/jev-recruiter/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z skeptrunedev

A Jev-driven recruiting agent with a local web UI that shows evidence quotes behind each decision; a short demo video is linked.

<p>A Jev-driven recruiting agent with a local web UI that shows evidence quotes behind each decision; a short demo video is linked.</p><p><strong>author demo:</strong> Demo.</p><p><a href="https://github.com/skeptrunedev/jev-recruiter">Open repository</a></p> https://jevland.mnimiy.com/projects/x-scanner-extension/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z oso95

A Chrome extension that labels posts in the X timeline with Jev across five dimensions while scrolling.

<p>A Chrome extension that labels posts in the X timeline with Jev across five dimensions while scrolling.</p><p><strong>author demo:</strong> $0.000033 per post, 179 ms average, about 80 posts and $0.0027 per minute of scrolling.</p><p><a href="https://github.com/oso95/x-scanner">Open repository</a></p> https://jevland.mnimiy.com/projects/ruby-decision-model-gem/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z Obie Fernandez

A zero-dependency Ruby client for decision models supporting TypeSafe's native API and OpenRouter, with retries, timeouts and usage/cost fields.

<p>A zero-dependency Ruby client for decision models supporting TypeSafe&#39;s native API and OpenRouter, with retries, timeouts and usage/cost fields.</p><p><strong>source-backed:</strong> Community gem (v0.1.0); separate from ruby_llm-typesafe.</p><p><a href="https://github.com/obie/ruby_decision_model">Open repository</a></p> https://jevland.mnimiy.com/projects/jev-pruner/ 2026-09-20T00:00:00Z 2026-09-20T00:00:00Z tamaratran

A Claude Code and Codex CLI plugin that intercepts Bash output above roughly 10K estimated tokens, chunks it and asks Jev which lines are still needed, dropping progress logs and boilerplate.

<p>A Claude Code and Codex CLI plugin that intercepts Bash output above roughly 10K estimated tokens, chunks it and asks Jev which lines are still needed, dropping progress logs and boilerplate.</p><p><strong>source-backed:</strong> 200-line chunks, 30K-token batches. Same author as fast-jev-compaction.</p><p><a href="https://github.com/tamaratran/jev-pruner">Open repository</a></p>