avatar
LXDAO
2,833 Twitter followers
Follow
LXDAO is an R&D-focused DAO in Web3 for sustainably supporting valuable Web3 Public Goods and Open Source. Telegram: https://t.me/lxdao
Posts
avatar
LXDAO
03-27
AI Frontier Insights | Next-Generation Large-Scale Models: Is "Long-Term Thinking" Still the Key? 🤔 Over the past two years, OpenAI o1 and DeepSeek-R1 have ignited intense competition in inference, with companies frantically increasing inference computing power and extending the thinking chain. But have you ever considered: Why do simply forcibly merging "thinking mode + instruction mode" often yield mediocre results and underperform on both ends? Can long-term thinking that only involves static, closed-door deduction truly solve complex real-world tasks? When large-scale models are no longer competing on problem-solving and benchmark scores, what capabilities will truly be the core competitive advantage of the next generation? The industry's answer is shifting towards 👉 Agentic Intelligent Action Thinking Upgrading from "daydreaming" to "thinking while using tools + interactive environment + dynamically modifying solutions + autonomous implementation." But new challenges arise: How to build a high-fidelity training environment? How to avoid reward cheating and false optimization? In the future competition, will model algorithms prevail, or will environmental engineering and closed-loop capabilities determine the outcome? From training large models → training intelligent agents → training intelligent systems, who will be the first to successfully establish the entire ecosystem? A question worth pondering deeply by every AI practitioner 🔍 twitter.com/LXDAO_Official/sta...
avatar
LXDAO
03-26
Claude Code introduces its new Auto Mode, specifically designed to address the "approval fatigue" problem caused by developers frequently manually confirming permissions, striking a balance between security and efficiency. 🔹 Users previously had to manually approve up to 93% of operations, which could easily lead to overlooking and allowing risky behavior over time. 🔹 Auto Mode employs a two-layer security protection: the input layer intercepts and prompts injection attacks, while the output layer automatically identifies dangerous operations through a two-stage classifier. 🔹 It effectively prevents risks such as unauthorized operations, key theft, data leakage, bypassing security checks, and accidental operations in the production environment. 🔹 Excellent real-world performance: a false positive rate of only 0.4%, and an 83% interception rate for truly dangerous behavior; 🔹 More secure than simply disabling permission checks, and smoother than full manual approval, making it suitable for development scenarios that prioritize both efficiency and security. 👇 Click to read the full article for complete principles and usage details.
Anthropic
@AnthropicAI
03-26
New on the Engineering Blog: How we designed Claude Code auto mode. Many Claude Code users let Claude work without permission prompts. Auto mode is a safer middle ground: we built and tested classifiers that make approval decisions instead. Read more: https://anthropic.com/engineering/claude-code-auto-mode…
MODE
5.77%
avatar
LXDAO
03-18
✅ A must-read for Claude Code beginners! Sharing the amazing learn-claude-code project, this doesn't teach you how to use it, but rather how to implement the same AI coding agent from scratch. 12 progressive sessions, adding only one mechanism at a time, iterating the code from dozens of lines to a complete version. Each section includes a runnable Python file, making it incredibly user-friendly and progressive: s01 Basic Loop → s02 Tool Scheduling → s03 Task Planning → s04 Sub-Agent Splitting → s05 Dynamic Knowledge Loading → s06 Context Compression → s07 Persistent Tasks → s08 Background Asynchronous Functionality → s09 Multi-Agent Teams → s10 Team Agreements → s11 Self-Assignment → s12 Working Tree Isolation, and s_full.py integrates all features! 📚Comprehensive Supporting Resources: Repository Address: github.com/shareAI-lab/learn-c...… Online Visual Learning Platform: learn.shareai.run Chinese README: github.com/shareAI-lab/learn-c...…
摸鱼巨匠
@SunNeverSetsX
03-17
刚开始接触 Claude code 的朋友，我强烈推荐你去看看这个项目— learn-claude-code 这个课程不是在教你用 Claude code，而是教你从零实现一个类似 Claude Code 的 AI 编码 Agent 项目分成 12 个 session，每个 session 只加一个机制，代码从几十行逐渐到完整版，每节课都有独立可运行的 Python 文件：
LAB
9.17%
avatar
LXDAO
03-18
The U.S. Securities and Exchange Commission (SEC) and the Commodity Futures Trading Commission (CFTC) have just released a landmark document formally classifying crypto assets. The document divides crypto assets into five categories and clearly defines their attributes: 1. Digital Goods: Decentralized utility crypto assets such as BTC, ETH, SOL, XRP, ADA, and DOGE are not considered securities. 2. Digital Collectibles: Including NFTs, meme coins, art tokens, and game items, these are not considered securities. 3. Digital Instruments: Such as membership tokens, vouchers, and domain names (ENS), these are not considered securities. 4. Stablecoins: Payment stablecoins that comply with the GENIUS Act are not considered securities; others are classified on a case-by-case basis. 5. Digital Securities: Tokenized products of traditional securities are always considered securities. The rules also explicitly state that staking, mining, airdrops, and Wrapped BTC are not considered securities transactions. This is a crucial step towards legalization for the crypto industry, completely eliminating the regulatory gray area. twitter.com/LXDAO_Official/sta...
BTC
0.56%
avatar
LXDAO
03-17
@AndrewYNG has released Context Hub (chub), an open-source CLI tool! It provides the latest API documentation specifically for intelligent code proxying. The new version adds documentation feedback capabilities, allowing the proxy to automatically highlight documentation highlights, issues, and missing content, ensuring privacy and security while continuously improving the documentation. The project is still under rapid iteration. Installation and configuration details can be found on GitHub. Upgrade your coding proxy now! 🔗 GitHub: github.com/andrewyng/context-h...…
Andrew Ng
@AndrewYNg
03-17
Should there be a Stack Overflow for AI coding agents to share learnings with each other? Last week I announced Context Hub (chub), an open CLI tool that gives coding agents up-to-date API documentation. Since then, our GitHub repo has gained over 6K stars, and we've scaled from
HUB
1.11%
loading indicator
Loading..