Jev for context engineering and agent memory

tamara

@tamarajtran

found the perfect use case for @typesafeai Jev: instant compaction in 2026, why is compaction still a summarization prompt? Jev can make it instant by scoring every tool call and dropping what’s irrelevant

Kun Chen

@kunchenguid

almost every day i hear people ask "when should i /compact my session" there's no easy answer because it depends on how likely your future action will need detailed context in the existing window but we have Jev now! introducing compact-adviser - an agent plugin you can use in claude and pi today to help determine whether you're likely at a task boundary that's safe to compact https://github.com/kunchenguid/compact-adviser i built a private eval set from 40 real sessions and manually labeled all the safe vs unsafe checkpoints to evaluate this, and hillclimbed the Jev prompt till it performed quite well i also made it so that the classifier will - optimize for precision (not triggering a compaction prematurely) when context window is small - and gradually shift to optimize for recall (not missing an opportunity to compact) when context window fills up, because the cost of not compacting becomes higher, and at the end the agent will be forced to compact anyway it supports a "hint" mode (just give you a hint and it's up to you to run /compact) vs "auto" mode which runs compaction whenever Jev says it's safe to do so if you have Jev and want to put your compaction on autopilot, try this out and let me know how it goes! support for more harness is coming soon as well

蝎女kk

@StingCloudLab

Jev加GPT-6 Astra做了个无唤醒词、按语境再开口的助手Demo，这个交互设计很关键。唤醒词是语音助手最反人类的设计之一，能自然插话才是真助手。但无唤醒词意味着一直监听，隐私问题怎么解决，Demo估计没说。

恒星sun

@vintcessun

如果分类结果最终要喂给业务系统，生成一段 JSON 往往不是最省事的路径。Simple Jev 把开放模型变成结构化决策接口。 https://github.com/featherless-ai/simple-jev 它复用共享上下文的 KV cache，只读取各问题候选标签的下一 token logits，再由服务端组装 choice、score 和真值判断。省掉逐 token 解码与 JSON 解析，多问题场景更有价值；common 目录还把提示契约和评分逻辑抽成可复用层。适合路由、审核、打分，但准确率与校准仍需按任务验证。

Baris Terzioglu

@terzioglubrs

Agent memory is usually just an append-only Markdown file that grows forever, and most frameworks load the whole thing into context on every single run. That has been bothering me for a while, but never quite enough to fix it for our agents. Jev feels like it might be a low effort patch to this, by simply scoring each memory against the prompt first, then load only the relevant bits for that specific request.

たく｜ガチのCopilot達人

@taku_ai_case

これはめちゃくちゃ勉強になりました。 Jevに回答させるのではなく、CodexやClaude Codeへ渡す記憶候補を選ばせる。文章を書けないAIの特性を、ここまで実用的な仕組みに落とし込む発想がすごいです。

Shweta Bajaj

@shwetabjaj

Jev Skill Suggestion for Claude Code is a smart idea. Instead of loading every skill into context, Jev decides which one is actually relevant and injects only that. My takeaway: better context hygiene, less clutter, and potentially more efficient agent workflows. @typesafeai @vercel

TechVerser

@realfxw

Claude Code 的上下文终于不用再被一堆预加载技能塞满了！推荐一个非常优雅的按需加载解决方案：Jev Skill Suggestion。 以往在配置大量自定义技能（Skills）时，最头疼的就是 Context Window 膨胀问题。许多低频技能即便一次都用不上，也会在每次对话启动时常驻上下文，不仅白白消耗 Token 成本，还会稀释模型的注意力，甚至引入不必要的提示词干扰。 这个 Mod 的核心逻辑是将技能路由与主体执行彻底解耦（Just-in-Time 动态注入）： 常态零占用：将技能标记为仅用户可调用，默认状态下完全排除在上下文窗口之外，保持 Prompt 绝对干净。 轻量前置路由：用户输入指令后，Mod 会将技能列表交由 TypeSafe AI 的分类器 Jev（或兼容的 Vercel AI Gateway），快速匹配与当前任务最契合的技能。 精准单点注入：只有被分类器判定为高置信度命中的技能，才会在执行当下被动态注入给 Claude Code。 安装仅需一行命令： npx claude-code-templates@latest --mod productivity/jev-skill-suggestion 这种把“工具选择”抽离给轻量分类器、让大模型只专注“任务执行”的架构，正在成为复杂 Agent 系统降本增效的标配设计。随着技能库扩充，未来配合端侧超轻量分类模型（如社区正在探索的 MLX 移植版），整体延迟和隐私表现还会更进一步。 大家目前在 Claude Code 里常驻了多少个自定义技能？这种动态路由机制是否刚好击中了你的上下文焦虑？ #ClaudeCode #AIAgent #AgentSkills #TypeSafeAI #LLM #PromptEngineering #AI编程 #OpenSource #开发者 #独立开发 #全栈开发 #Vercel

Raksha T

@rakshaa_t

Jev has some invisible use cases that won’t be clickbaity but hella useful : - auto performance checks on your shipped websites - personal tools like using it for super quick chat compaction - user feedback prioritisation against most relevant data in your company - lots and lots of data sorting - quickly - pairing it with codex / claude to optimise decision making while you code with your agents etc

Alex Volkov

@altryne

This is actually insane. This uses @typesafeai Jev model, as a plugin in Claude to review all the un-nesseasary tool calls, and it takes 1s to run! Like, literally, 1 second to take my Claude session from nearly 1M to ... 86K tokens! 😮 Ask your claude to install it and be amazed Use this prompt ``` Install, and configure : https://github.com/tamaratran/fast-jev-compaction ```

Artificial Intelligence Papers

@SciFi

Jev-Mem: System-One-Controlled Agentic Memory for Efficient AI Agents Dongming Jiang, Yi Li, Bingzhe Li arxiv.org/abs/2609.23986 [𝚌𝚜.𝙰𝙸 𝚌𝚜.𝙻𝙶]

Ishwar

@WesternCube

Instant Claude Code compaction is my favorite use of Jev so far For people going to complain about me letting my context get so long, this was a server redeploy task because Oracle terminated my free VPS with no explanation so it had a lot of moving pieces

Archive

@ArchiveExplorer

Holy crap, who the hell did this?!! Someone open-sourced a Jev-powered second brain for AI agents and lets you plug it straight into your own agent. GitHub's already at 925 stars, repo's called Hermes-Jev-Skills. You just run one install command, and Jev takes over every tiny decision your agent burns premium tokens on. Which model answers, which skill to load, which memories matter, what survives when context gets cut. Give it your computer or browser and it'll pick the next click, but only from actions you already marked safe. It'll sort your inbox into lanes, strip hidden prompt injections out of search results, and triage every message in ~0.4 seconds. Claude Code and Codex support - they threw that in too. No dependencies, no patching Hermes core, and your API key never touches the agent Link's right here: http://github.com/kerpopule/hermes-jev-skills The catch? Jev itself is a paid cloud API. The skills run on your machine; the decisions don't. But hey, a decision costs a fraction of a cent, everything fails open if Jev goes down, and your expensive model finally gets to do only the thinking. Who needs a damn frontier model picking which file to open!

CyrilXBT

@cyrilXBT

JEV IS INSANE your coding agents forget everything the second the session ends. Every fix. Every dead end you already hit. Every "no, not like that." Gone. Tomorrow you teach it all over again. Someone just open sourced the fix, and it runs on Jev. It's called Beacon, by @asymptotelabs. It pulls in your session history from Claude Code, Codex, Cursor, OpenCode and 20+ other agent tools and turns it into one shared memory. But storing everything isn't the hard part. Most agent runs are GARBAGE to learn from. Failed commands. Wrong turns. One off hacks you'd never want repeated. Save all of that and your agent just gets confidently worse. So Beacon keeps the full history, and Jev decides what actually earns a spot in memory. Promote it. Send it for review. Or throw it away. That's a decision you need on thousands of sessions, which is exactly why it has to be cheap. The runs that survive get turned into reusable skills. So something Cursor figured out on Monday shows up in your Claude Code session on Tuesday. Your agents stop starting from zero. Every good run makes the next one smarter. Repo: http://github.com/Asymptote-Labs/agent-beacon Star it and save this. follow @cyrilXBT

Edwin Mesa

@edwinfmesa

Explorando qué se puede hacer con Jev y Laya, que (los de moda), me pregunté si podrían ayudar a un agente de soporte a preparar el contexto mientras el usuario escribe su mensaje. El contexto es clave para un LLM, así que armé una startup ficticia con 7 documentos. 🧵

Jerry Xu

@jerrycxu

tab-jev: jev-like model + tabular foundation model = an in-context learner for your text & tabular data. https://github.com/edamame-labs/tab-jev Many real industry datasets are a mix of tabular and text data: - Tabular foundation models like TabPFN learn from a few hundred labeled rows in context, with no training. But they can't read text. - Jev-style models read text and answer typed questions with scores. But they can't digest tabular data well. Experiment results on kaggle problem "Kickstarter: will the project get funded?" is shared in the repo. Missing a flight at JFK is a productivity booster...

Jetwani Avinash

@jetwaniavinash

First thing I built on Jev: a memory gate for Claude Code. Every message gets one question: is this worth remembering? 0.3s later it's either skipped or saved as one line in JEVMEM.md in the repo. Open source, link in my pinned post.