标签：#ens

FROST 的治理哲学：为什么每个 Agent 系统都需要一部宪法 "细胞会死，但谱系会存续。Agent 会消亡，但宪法会传承。资产会永存。" 在 AI Agent 框架百花齐放的 2026 年，我们见证了一个有趣的现象：框架越来越多，治理却越来越缺失。LangChain 给了你工具链，CrewAI 给了你角色扮演，AutoGen 给了你对话——但没有人回答一个关键问题： 当你的 Agent 家族有十代、百代，谁来保证它们的行为不越界？ 这就是 FROST 要解决的问题。 一、一个被忽视的问题：Agent 治理真空 想象这样一个场景：你构建了一个 Agent 系统，它帮你处理客户咨询、自动写报告、管理财务。一开始只有 3 个 Agent，你都能监控。三个月后，Agent 数量变成了 30 个——有些是你创建的，有些是 Agent 自己创建的。 问题来了： 某个 Agent 开始访问不该访问的数据 某个 Agent 创建的子 Agent 继承了一份不该继承的权限 某个 SOP 工作流在运行中被人悄悄修改了 这不是科幻场景，这是今天 Agent 系统正在发生的事。 根本原因 ：现有框架的"治理"要么是提示词级别的软约束（"请你不要访问这些数据"），要么是事后的日志审计。没有一个是代码级别、架构级别的硬约束。 FROST 的设计哲学是： 治理必须是架构的一部分，而不是补丁。 二、FROST 的宪法模型：四层治理 FROST 把治理拆解为四个层次，每一层都有对应的代码实现： 第一层：只读继承——权限的代码级强制 在大多数框架中，Agent 之间的数据共享靠的是"约定"。FROST 用的是 HierarchicalStore ——一种层级化记忆容器： class HierarchicalStore : """ 层级化记忆存储。 祖先 Store 对后代只读——后代可以继承，但不能篡改。 """ def __init__ ( self , data = None , parent = None , readonly = False ): self . _data = data or {} self . _parent = parent self . _readonly = readonly # 关键：只读标记 def save ( self , key , value ): if self . _readonly : raise PermissionError ( " 只读 Store 禁止写入 " ) self . _data [ key ] = value def load ( self , key , default = None ): if key in self . _data : return self . _data [ key ] if self . _parent : return self . _parent . load ( key , default ) return default 注意 readonly 参数。当一个 Agent 创建子 Agent 时，子 Agent 对父 Agent 的 Store 是 只读的 。这不是提示词告诉 Agent "请你不要修改"，而是代码直接抛出异常。 这意味着 ：即使 LLM 产生了幻觉，试图越权写入，代码层面也会拒绝执行。 第二层：SOP 宪法校验——工作流的可审计性 在 FROST 中，SOP 不是随便定义的步骤列表，而是经过祖辈审核的"宪法文本"： def validate_sop ( ancestor_sop , descendant_sop ): """ 祖辈验证后代的 SOP 是否合规。 检查点： 1. 后代 SOP 的步骤不能超出祖辈定义的边界 2. 后代 SOP 不能调用未授权的 Skill 3. 后代 SOP 的数据访问不能超出权限范围 """ # 检查 1：步骤边界 for step in descendant_sop . steps : if step not in ancestor_sop . allowed_steps : raise SOPValidationError ( f " 步骤 { step } 超出祖辈定义的边界 " ) # 检查 2：Skill 授权 for skill in descendant_sop . required_skills : if skill not in ancestor_sop . authorized_skills : raise SOPValidationError ( f " Skill { skill } 未获授权 " ) return True 这意味着 ：任何后代 Agent 想要执行的工作流，都必

Last week, I asked my AI agent to scrape 300 job listings from a recruiting site. It broke at row...

# Sakana Fugu: The Multi-Agent AI System That Works Like a Team We’ve all been there: copy-pasting a prompt from ChatGPT to Claude, and then to Gemini, trying to find which AI gives the best answer. Different AI models have different strengths. Some are excellent programmers, some write beautifully, and others are master logicians. But constantly switching between them is slow and frustrating. Sakana AI has introduced a brilliant solution: Sakana Fugu . Fugu is a multi-agent AI system that bundles multiple frontier models into a single, seamless package. To you, it looks like a single chatbot. But behind the scenes, it acts as a project manager, coordinating a team of top-tier AI models to solve your task. What is Sakana Fugu? Sakana Fugu is a collaborative AI framework. Instead of relying on one massive AI model to do all the work, Fugu orchestrates a pool of specialized models (such as GPT-5, Claude Opus, and Gemini Pro). When you give Fugu a complex, multi-step prompt, it: Analyzes the task and breaks it down into steps. Assigns specialized roles (like researcher, coder, and editor) to different AI models. Passes the work back and forth between them until the final, polished result is ready. By working as a team, these models achieve far better results than any single AI could on its own. The Secret Sauce: Teamwork Over Size Fugu’s intelligent coordination is based on two core concepts presented at the ICLR 2026 conference: 1. The Manager (TRINITY) Think of TRINITY as the manager of the team. It is a compact coordinator model that assigns specific roles to the worker LLMs. For example, it might tell Gemini to generate the initial code, tell Claude to find the bugs, and tell GPT to write the user documentation. They collaborate seamlessly without needing to merge their code bases. 2. The Conductor The Conductor acts as the communication network designer. It figures out the best way for the worker AIs to talk to each other for your specific question. It writes cust

Introduction Although we interact with LLMs using natural language, these models never processes raw text directly. Before a prompt reaches the model, it is converted into a sequence of tokens , the fundamental units that the model understands. Tokenization is one of the earliest stages of the inference pipeline and influences everything from context windows and API pricing to latency and memory usage. What Is a Token? A token is the smallest unit of text processed by a language model, it is not necessarily a word. Depending on the tokenizer, a token may represent: an entire word part of a word punctuation whitespace numbers symbols emojis Different models use different tokenizers, so the same text may be split differently depending on the model. Why Tokens? Simply because language models operate on numbers, not text. Before the transformer can perform any computation, the input must be converted into a numerical representation. The preprocessing pipeline looks like this: Raw Text │ ▼ Tokenizer │ ▼ Tokens │ ▼ Token IDs │ ▼ Embedding Layer │ ▼ Embedding Vectors │ ▼ Transformer The tokenizer splits the input into tokens and each token is then mapped to a unique integer called a token ID , which are passed through the model's embedding layer, which converts them into dense vectors that become the actual input to the transformer. A Real Example Instead of using hypothetical examples, let's look at how OpenAI's tokenizer processes text. Input: I have no enemies. OpenAI tokenizes it to: ["I", " have", " no", " enemies", "."] with the following token IDs: [40, 679, 860, 33974, 13] that have been generated by OpenAI Tokenizer for the "GPT-5.x & O1/3" models. The transformer never sees the original sentence, it only receives the corresponding sequence of token IDs. Token IDs After tokenization, every token is replaced with an integer. Conceptually: " have" → 679 " no" → 860 " enemies" → 33974 ... The exact numbers differ between models because each tokenizer has its own voca

🚀 CNTRL by Omnikon Org Selected for Elite Coders Summer of Code (ECSoC) 2026 Building an AI-first browser for developers—and taking the next step through ECSoC 2026. Open source has always been one of the best ways to learn, collaborate, and build software that makes a difference. Today, I'm excited to share a milestone that means a lot to our team. Our project CNTRL , developed by Omnikon Org , has officially been selected for Elite Coders Summer of Code (ECSoC) 2026 ! 🎉 This selection gives us an incredible opportunity to collaborate with contributors worldwide and continue building a browser that's designed from the ground up for developers. 💡 Why We Started CNTRL Every developer has experienced this workflow: Open documentation Search GitHub Ask an AI assistant Open Stack Overflow Copy code Switch back to the IDE Repeat... The browser has become the center of development, but it still isn't designed for developers. We wanted to change that. Instead of building another browser, we started building one where AI is part of the experience—not another tab. 🌐 What is CNTRL? CNTRL is an AI-powered browser built for developers. Our vision is to create a browser that understands how developers work and helps them stay focused. Some of the ideas we're working toward include: 🤖 AI-assisted coding 📖 Context-aware documentation 💬 Built-in developer assistant ⚡ Faster research workflows 🔌 Extensible architecture 🌍 Community-driven open source The project is still evolving, and ECSoC gives us the perfect platform to accelerate its development. 🏆 Selected for ECSoC 2026 Being selected for Elite Coders Summer of Code 2026 is a huge milestone for our organization. It means we'll have the opportunity to: Collaborate with talented contributors Improve the project's architecture Build exciting new features Learn from the community Grow CNTRL into an even better developer tool We're incredibly thankful to the ECSoC team for believing in our vision. 🌍 About Omnikon Org Omnikon Org is

Hey devs 👋 OrinIDE is an AI-powered code editor that runs entirely in your browser — no...

There's a pattern playing out across indie hacker forums, engineering blogs, and Discord servers right now: a founder builds a product on GPT-4-class models, ships it, gets traction — and then opens a bill that makes them question the whole business model. LLM costs have a nasty habit of scaling linearly (or worse) with usage, right at the moment a product starts succeeding. In response, a growing body of developer tutorials is focused on one goal: keeping the intelligence, dropping the invoice. Think of it like the early days of cloud hosting. Companies over-provisioned expensive dedicated servers until autoscaling and cheaper commodity infrastructure made "pay for what you use, and use less of the expensive stuff" the default architecture. The LLM ecosystem is going through its own version of that shift right now, and DeepSeek has become the poster child for the "just as capable, dramatically cheaper" alternative to OpenAI's premium models. Migrating With Minimal Friction The first core question developers are wrestling with is deceptively simple: how do you swap out a model provider without rewriting your whole application? The answer that keeps surfacing is API compatibility layers . Many cost-effective providers, including DeepSeek, expose an API that mirrors OpenAI's own request/response format almost exactly. That means in a lot of codebases, migration isn't a rewrite — it's a find-and-replace of a base URL and an API key. # Before: pointed at OpenAI client = OpenAI ( api_key = " sk-openai-... " , ) # After: same SDK, same code, different provider client = OpenAI ( api_key = " sk-deepseek-... " , base_url = " https://api.deepseek.com/v1 " ) That's it, in the simplest case. Because the OpenAI Python SDK just talks HTTP under the hood, any provider that speaks the same "dialect" can slot in without touching your prompt logic, your function-calling schemas, or your downstream parsing code. The real friction shows up in the details: subtle differences in how mode

The right desk gadgets can help you reduce clutter, stay focused, and add a little extra convenience to your day.

Bending Spoons remains largely unknown, even as its portfolio of products has served more than a billion people.

GitHub热门项目 | Macro is a unified interface for email, messages, tasks, calls, agents, pull requests, docs, crm — linked together with shared AI memory. | Stars: 366 | 67 stars this week | 语言: Rust

GitHub热门项目 | A new type of shell | Stars: 39,869 | 8 stars today | 语言: Rust

GitHub热门项目 | Cyber Security ALL-IN-ONE Platform | Stars: 7,499 | 31 stars today | 语言: TypeScript

GitHub热门项目 | 📨 The ultimate agentic social media scheduling tool 🤖 | Stars: 32,727 | 38 stars today | 语言: TypeScript

GitHub热门项目 | The fastest knowledge base for growing teams. Beautiful, realtime collaborative, feature packed, and markdown compatible. | Stars: 39,409 | 69 stars today | 语言: TypeScript

GitHub热门项目 | Give Claude the ability to watch any video. /watch downloads, extracts frames, transcribes, hands it all to Claude. | Stars: 3,377 | 356 stars today | 语言: Python

GitHub热门项目 | Persistent file-based planning for AI coding agents and long-running agentic tasks. Crash-proof markdown plans that survive context loss and /clear, plus a deterministic completion gate and multi-agent shared state on disk. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard. | Stars: 24,557 | 61 stars today | 语言: Python

We just released an important update: retraining our Random Forest model on real malware behavior from the CIC-MalMem-2022 dataset. The challenge was mapping 58,000 complex memory dump traces into a clean 10-feature vector space that our syscall graph extractor produces. How it works: Sandbox target in Docker (network dropped) Trace every syscall with strace -t -f Parse into a NetworkX directed graph Extract 10 features (process count, network connections, file operations, severity scores, etc.) Feed into RandomForest for classification We also resolved module-level import cycles and pinned skops for safer model deserialization in production. Looking for collaborators who understand malware behavior, syscall parsing, or want to contribute detection rules. Open to issues and PRs. https://github.com/tejasprasad2008-afk/TraceTree

Your order-processing Lambda starts sending duplicate confirmation emails. Not always — maybe one order in twenty. CloudWatch shows more invocations than messages published. The function code hasn't changed in weeks. What changed is that someone added a fraud check that pushed processing time from 25 seconds to around 45, and your SQS queue is still running the default 30-second visibility timeout. That combination is the whole bug. When a Lambda pulls a message from SQS, the message isn't deleted — it's hidden for the duration of the visibility timeout. If the function is still working when that window closes, SQS assumes the consumer died and hands the same message to another invocation. Now two Lambdas are processing the same order, both will "succeed," and both will send the email. Nothing errors. Nothing retries. There is no log line that says "this message was delivered twice because your timeouts are misconfigured." Infrawise ( npm ) flags this exact mismatch as a high-severity finding before it costs you an afternoon of staring at idempotency-free handler code. This post walks through why the bug is so hard to see, how the detection works, and how to keep an AI assistant from reintroducing it. Why you never catch this one yourself Three things make this misconfiguration nearly invisible: It passes every test. In local tests and staging, your handler processes a synthetic message in two seconds. The 30-second visibility window never comes close to expiring. The bug only exists under production conditions — real payload sizes, real downstream latency, cold starts stacking on top of slow dependencies. The defaults set the trap. SQS queues default to a 30-second visibility timeout. Lambda functions routinely get their timeout bumped to 60, 120, or 900 seconds as they grow. Nobody bumps the queue at the same time, because the two settings live in different consoles, different IaC resources, and usually different pull requests. The failure signature points elsewhe

The problem with a data breach is not finding evidence. It is connecting it. But let me start where I actually was: 4 AM, last day of the hackathon, staring at this in my terminal. RateLimitError: GroqException - Rate limit reached for model `llama-3.3-70b-versatile` on tokens per day (TPD): Limit 100000, Used 99787, Requested 1616. Please try again in 20m12s. Used 99,787 out of 100,000. My deployment was half done, my demo graph was empty on the server, and the free tier had 213 tokens left. The submission deadline was hours away. I had not slept. I had not eaten. My friends were asleep and I was swapping API keys like a gambler swapping chips. This post is the story of how we got there, and how it ended at 7 in the morning with the best sigh of relief I have ever taken. First, some honesty about how I got here When I joined my first WeMakeDevs hackathon, I did not believe in it. I thought it was one of those ordinary online events. Fake prizes, no follow-through, what would I even get out of it. I joined anyway, mostly out of boredom, got into the Discord, talked to people, made a few connections. I landed in the top 50. A few days later an email showed up: a free Claude Max subscription as a gift. I read it twice. I genuinely could not believe a hackathon had actually delivered something. So when this hackathon opened, I did not hesitate. I messaged my friends and said we are joining as a team this time. Three of us: me (Mehraan), Aqib, and Ubaid. The spark We spent the first evening in our group chat throwing ideas around and shooting most of them down. Then one of my friends dropped a thought that stuck: what happens after a company gets hacked? I started digging into it. The answer is honestly depressing. After a breach, the evidence is everywhere. VPN records. File access logs. The email gateway. Badge readers at the office doors. CCTV. HR notes. Anonymous tips. Each system tells one small piece of the story, and a human analyst has to stitch all of it togeth

Have you ever noticed this? You explain your project to an AI chatbot, have a great conversation, then come back later... and it asks you to explain everything again. Start a new chat, and it's like meeting you for the first time. This isn't a bug. It's how most AI models work—they don't remember past conversations on their own. That's where Cognee comes in. Instead of making AI start from scratch every time, it gives AI a way to remember what matters. Why does AI forget everything? Most AI models don't have long-term memory. Every time you start a new chat, the AI only knows what you send in that conversation. It doesn't remember your previous chats, your project, or your preferences unless you provide them again. A common solution is RAG (Retrieval-Augmented Generation) . It stores your documents in a searchable database so the AI can look up relevant information when needed. RAG genuinely helps, but it only knows what sounds similar. It doesn't know "Priya" and "the payments lead" are the same person, or that this week's ticket shares a root cause with one from March. Similarity search finds neighbors — it doesn't understand relationships. That's where Cognee takes a different approach. What is Cognee? Cognee is an open-source memory layer for AI applications. Instead of making an AI start from scratch every time, Cognee helps it remember information across conversations. It can learn from your documents, files, websites, or notes and use that knowledge whenever it's needed. Unlike traditional RAG systems that mainly find similar text, Cognee also understands how different pieces of information are connected. That gives AI more accurate and meaningful answers. It's Apache-2.0, runs locally by default, and has 27k+ GitHub stars. How does it work? At a high level, the process is simple: flowchart LR A[Text, Files, URLs] --> B[remember()] B --> C[Cognee builds AI memory] C --> D[recall()] D --> E[AI answers using remembered knowledge] For example: "Alice bought a Pr