Jev for coding agents — a daily feed

Coding agents, linters, SDKs and plugins for Claude Code and Codex that people built with Jev.

253 builds
Stanislav Sorokin@stas_sorokin_ yesterday
Audited 1,000 GitHub repos for CLAUDE.md and AGENTS.md fossils
JEV just audited the CLAUDE.md and AGENTS.md files of GitHub's 1,000 most-starred repos for $0.02 🤯 Claude Opus 5.5 on the same 609 flagged lines, same questions: $0.81, and 10x slower. What prompt-fossils hands back: → the lines that hobble Opus 5.5, with file and line number → which dated pattern each one is: shouting, bare "never" lists, step scripts, fossils → which rules are load-bearing and
$0.02$0.8110× faster
Coding agents 795 views
Takashi Minoda@aad34210 yesterday
Jev API connected to Dataiku Python Recipe
#Dataiku にPython Recipeで実行をさせたJevとのAPI連携の最初のファーストステップができた。 回答も同じなのを確認したので、JevのAPIキーを使ってPython Recipeで動くことまで確認ができた。 https://t.co/yO9a3HbjEc
Coding agents 85 views
うめぼし@UmeboshiCenter yesterday
DJ interruption system connected to a local Jev
友達がハエ🪰(ハエ?!w)が DJ中にちゃちゃ入れてくるマギシステムのダミーリポジトリをくれた🪰 自分のjevと繋げたらもう普通に動いてる!!!😂🪰 ダミーにしては出来すぎてるw本番 繋いでみます🎵 https://t.co/fcAKcXz6Po
Coding agents 83 views
無胥 | Starknows@starknows99 yesterday
Test log article written with Jev
這幾天玩了一下 jev 請 AI 幫我寫了一篇測試紀錄 https://t.co/mIHqgjHtTe
Coding agents 76 views
Takumi KOSHII@takumi_koshii yesterday
Automation script selection task ran with Jev, 52% of Codex time
Jevのユースケースをあれこれ考えていたのですが、試しにこれを試してみたら思ったより効果があったので記事にしてみました 定型作業の自動化スクリプトを選ぶタスクを Jev に任せたら Codex の 52% の時間でできた話 #DevelopersIO https://t.co/sIXaJ36JXN
Coding agents 28 views
Wayne Radinsky@waynerad yesterday
jev-lint static analyzer for JavaScript and TypeScript
jev-lint (no capitalization) is a static analyzer-ish program that analyzes source code -- only JavaScript and TypeScript -- against a document in English of conventions the code has to follow, and generates error messages that are intended for your code-generating AI agent to read. As the name implies, it uses Jev. https://t.co/RoJrMHlDym #solidstatelife #ai #genai #llms #codingai #rlcd #jev #sta
Coding agents 23 views
Kushtrim Visoka@kushtrimvisoka yesterday
Reproduced Amnesty judgment coding on 199 Kosovo cases, 91% match
Can Jev reproduce @amnesty's coding of domestic violence judgments? I gave Jev (by @typesafeai) Amnesty's 199 Kosovo judgments, all in Albanian, and its definitions. It matched Amnesty's human researcher on 91% of 12,869 answers, and 97% on prison terms and fines. https://t.co/jPQkNSbuqb
91% accurate97% accurate
Coding agents 23 views
Ram Vinjamuri@RamV2003 yesterday
Judge benchmark: 235ms and $0.0059 for 360 calls
(5/5) none of this makes jev useless. it's still the fastest, cheapest judge i've run (235ms, $0.0059 for 360 calls, matched sonnet on sabotaged diffs at ~1/50th the cost). it just has one rule: the option set has to contain the truth how i'm running it now: 1. put the true answer in the options ("same", "neither", "not enough info"). 24/24 on the sabotage set in one call 2. ask twice, swap positi
235 ms$0.005950× faster
Coding agents 20 views
Dai@daioio yesterday
Implementation plan and PR verifier for tickets and specs
Finding @typesafeai Jev really useful for verifying implementation plans / tickets are ready to work on. And when a PR is opened verifying they match the spec and plan along with other criteria. Seeing where else I can slot this into my workflows but so far loving it 👏 https://t.co/BdVFz0sFjt
Coding agents 18 views
Mada まだ早い@madaearly yesterday
Measured 20 popular Jev repos and where they call it
Nineteen of the twenty most-shared Jev repos call it from their product code. The claim going around is that ninety percent of them are larping. I cloned all twenty and ran the same count on each. 20 of 20 reference Jev somewhere in source 19 of 20 do it in product code, not just tests or scripts 13 of 20 have a test that touches it median share of the codebase: 54.1% twelve of twenty are over hal
20 items19 items13 items
Coding agents 17 views
erKam 🅰️@erkamyaman_ng yesterday
jev-enforce CLI check for code, with benchmark and README
repo: https://t.co/JOAbTibMTU npx jev-enforce check --as code src/orders.ts no key = it does nothing, so it can't break your session. benchmark + every mistake it made is in the README
Coding agents 2 views
Denis Yarats@denisyarats 2 days ago
Autonomous Jev-like model trained with a swarm of agents
fun weekend project: AutoJev. i was curious to see if i could train a competitive Jev-like model completely autonomously with a swarm of agents using our internal system. turns out you can get quite far! some details: - gave the swarm a devbox with an h200 gpu - the swarm is a mix of astra and fable; used sol and luna for synthetic data and filtering; ran for 20 hours - spent $3.1k in total ($1.9k
Coding agents 15k views
かなめ|個人開発@BrainRevApp 2 days ago
GitHub Jev repo analysis: 3,645 repos classified into 22 buckets
GitHubのJev関連3,645件を、Jev自身に22分類させた。レポジトリが多い順に、 1. デモ・試作 531件 2. コーディングエージェント拡張 333件 3. 利用者向けアプリ 318件 14,044回の判定で費用は約0.78ドル。分類を眺めると、想像より用途の広さが出ている。 また、3位のアプリ向け用途が最も星を獲得している。 各カテゴリ内の分析結果は別ツイートにて。
3,645 items$0.78
Coding agents 7k views
いのま@iNoma_main 2 days ago
Kotlin library for Jev API type safety
「TypeSafe AI の API が type-safe じゃなかったので、KMP ライブラリを作った」 昨日公開したJevのKotlinライブラリについて、記事を公開しました! https://t.co/ynzw6sGyhs
Coding agents 4k views
Amodex@amodexbt 2 days ago
Coding loop with 32 steps and 158 decisions
JEV DOESN'T WRITE YOUR CODE. IT MAKES THE FIVE DECISIONS AROUND EVERY LINE OF IT. HERE'S THE LOOP every step your agent takes is really five judgment calls. which files to read. which model writes this. is this safe to run. keep this output or drop it. are we actually done. today one frontier model makes all five, and you pay frontier prices for every one. jev coding loop: 32 steps, 158 decisions,
32 items158 items7 items
Coding agents 2k views
Shota Oki@phensen123 2 days ago
Strands hook integration for Jev
JevをStrandsのフックに入れる話を書きました。脇道でStrands Shellをはじめて使いました、DockerなしでDockerのvolumeみたいなことができて便利 https://t.co/3KUVhb6HgT
Coding agents 1k views
avrl ☘@avrldotdev 2 days ago
Jelight code highlighter backed by Jev
Introducing Jelight v0.1.0 A code highlighter lib backed by Jev. Fully Open Source with no external dependencies. Works with both @vercel & @typesafeai api gateways. https://t.co/MliJSuL8GX
Coding agents 1k views
小畑タカユキ|AI×Web制作@大阪@no_ai_no_life 2 days ago
One-day personal tool built with Claude Code and Jev
「AI、何から始めればいいかわからない」という人に、僕が一番すすめたいのは、自分の困りごと用のツールを1個つくることです。 売るためでも誰かのためでもなく、まず自分のために。 決め手はコードが書けるかどうかじゃなく、「何を作るか」を自分の言葉で言えるかどうか。僕はそれをClaude CodeとJevで、1日で形にしました。サーバー代は月7.25ドル。 その手順を、初心者が上から順にやれば動くものが残る粒度でnoteに書きました。プログラミング未経験でも大丈夫です🔥 この投稿をリポストしてくれた方は、お得に読めます🙏 ↓ Claude Code×Jev 初心者が自分専用ツールを1日で作る12ステップ https://t.co/wLj7ebqzIo
$7.25
Coding agents 339 views
Yuepan Chao@smallnest 2 days ago
Local Laya wrapper for a Jev-compatible API
写了个程序,把本地 laya 包装成 jev 兼容的API https://t.co/7HJeWdPool
Coding agents 320 views
Nozomi Koborinai@fender_kn 2 days ago
Open source dev-flow improvement project using Jev
jev のキャッチアップも兼ねて、SDD をはじめとした開発フローの改善に繋がらないかなと思い、OSS を作ってみた 🙌(久しぶりの Zenn ブログとともに) https://t.co/kEmnPWiTpD
Coding agents 179 views
sat0xshi@sat0xshi 2 days ago
Added logging to a Grokbot powered by Jev
Grokbotに仕込んだJevにログを出す仕様を追加してみた。なるほど。 https://t.co/o8OwBFJVHz
Coding agents 165 views
Elixir Forum@elixirforum 2 days ago
Elixir client for Jev with TypeSafe and OpenRouter support
[Announcing] jev_elixir - elixir client for Jev with TypeSafe and OpenRouter support https://t.co/IyRRk23ENj #ElixirLang #MyElixirStatus
Coding agents 165 views
らむたき@電子工作@RamTuckey 2 days ago
Code-breaking experiment app with 48 Jev reviews
判断特化 AI「Jev」でコードを壊す実験アプリを作りました。 Next.js の Route Handler から Vercel AI Gateway 経由で 48回審査。 API キーをサーバー側に閉じる構成、実測とシミュレーターを混ぜない設計、CI ゲートの閾値とコスト試算まで。 #Jev #Nextjs #AI開発 https://t.co/uggZkzEsHZ
Coding agents 153 views
mu-cho243@mucho243 2 days ago
ServiceNow PDI connected to Jev via API
早速Qiita記事として公開しました (ノ・ω・)ノポイッ 新しい外部サービスとのAPI連携も実装できちゃうのだからBuild Agentさまさまである #BuildWithBuildAgentTokyo [ServiceNow] とりあえずServiceNow PDIとJevとをAPIで繋いでみた (Build Agent) https://t.co/Owe9Szyp5N #Qiita @mucho243より
Coding agents 150 views
Aynyan@BCNOFNe.sui 「 🦑 」@Aynyann 2 days ago
Jev voice assistant and Claude Code integration notes
Jev を音声アシスタントと Claude Code に入れた記録。 ・確認省略は「何を渡すか」で決まった ・自分で入れたゲートに自分が引っかかった ・流行りの圧縮プラグインは計測して見送り ・落ちた時に黙らん仕組みが一番効いた 全部 note に書いた https://t.co/K4C7wh7xKg #Jev #TypeSafe #ClaudeCode #AIエージェント
Coding agents 130 views
Nick · Raincoat@Raincoat_t 2 days ago
Open-sourced reusable Jev skills for coding harnesses
7 Jev skills you can reuse in Claude Code and Codex. I open-sourced Jev Skills for anyone building an AI coding harness with TypeSafe's Jev. The reusable pieces: 1. Model router: suggest the smallest suitable model from your available options. 2. Skill picker: find the relevant skill in your installed catalog. 3. Context picker: choose which file or excerpt to inspect next. 4. Test prioritizer: pi
Coding agents 116 views
Russ Wonsley@RussWonsley 2 days ago
Keyword rules vs Jev demo for message resolution
Same messages. Same planner. Keyword rules vs TypeSafe’s Jev. Small demo. Change the wording and watch what resolves. What would you throw at it first? https://t.co/7z9F9aGpwK https://t.co/reTf3rZavL
Coding agents 97 views
シープボーイ@Sheep_boy_game 2 days ago
Adjusted Jev system prompt for a conversational game
Jevに渡すシステムプロンプトを少し直してみました。 ただ今度はだいぶ甘口になりました。 Jev君は極端なんだよなぁ......。 https://t.co/8mLKNeyQsq
Coding agents 95 views
Akshay Subramaniam@AkshaySubr42403 2 days ago
Coding-agent rule checker, 0.2s p50 and $0.08 per 1k checks
Late hop on the Jev train! I built a hook that checks your coding agent against your team’s rules before you see its work. I compared it to GPT 6 so I'm not unoriginal. Jev: 0.2s p50 with $0.08 per 1k checks GPT-6: 2.6s p50 with $0.21 per 1k checks (but way better) https://t.co/rC0Lzk5IYP feel free to rip off of it!
0.2 s$0.082.6 s
Coding agents 79 views
ダニー|AIで人生デバッグ中@dadionai 2 days ago
DADIOS integration: 9 Jev runs with 6 successes and 1 timeout
JevをDADIOSに実際に入れてみたよ〜。 9回動かすと、成功6回・低信頼2回・タイムアウト1回。 面白かったのは、全部を任せるより『Codexの前後で小さな判断だけ先回り』させた方が、実務では使いやすかったこと。 何を任せ、どこは任せなかったのかまでまとめました👇 https://t.co/OE7plX4dsX
Coding agents 76 views
Alex Ollman@_ollman 2 days ago
Fine-tuned Laya on app data with 6x lower latency
I spent the weekend seeing if Laya, an open-source decision model equivalent to Jev, could be fine-tuned on an application-specific dataset and achieve equivalent results. It can, while cutting inference latency down by 6x. All fine tuned on a MacBook. https://t.co/1Yl5vjU4nI
6× faster
Coding agents 69 views
J.Gando (ジェイ)@jay34986 2 days ago
Dependabot PR auto-approval candidates with Jev
JevとGitHubのDependabotを組み合わせてみたブログを書きました。 JevでDependabot PRの自動承認候補を判定してみた|J.Gando https://t.co/F6sE76mnS6 #zenn
Coding agents 61 views
mu-cho243@mucho243 2 days ago
ServiceNow PDI connected to Jev by API at $0.0001 per request
とりあえずServiceNow PDIとJevとをAPIで繋いでみる <$0.0001 / 1 requests https://t.co/Z43v7U0Nl9
$0.0001
Coding agents 49 views
217 more below

Submit a build

Paste the link to your post on X about something you built with Jev. We check the link right away. The next update, within about 4 hours, adds the card if the post shows your build.