Today's brief

Hallmark 是一款开源 design skill，专为 Claude Code、Cursor 与 Codex 打造。它内置 57 道 slop-test 门控与 20 套主题系统，在生成前强制 AI 进行自我批判，从根本上拒绝千篇一律的「AI 味」模板，让不同 brief 产出气质迥异的站点；同时支持 audit、redesign、study 四种命令，可直接嵌入现有工作流。

TradingAgents 是一个基于 LLM 多智能体的开源金融交易框架，最新 v0.2.5 版新增了情感分析师、多模型覆盖（GPT-5.5 / Claude 4.6 / Qwen 等）及远程 Ollama 支持。虽然面向量化交易场景，但其角色分层的 Agent 架构与 LangGraph 状态管理实践，对设计复杂 AI 协作工作流有一定参考价值。

JetBrains 发布围绕 pi coding agent 构建的 agentic IDE ThinkRail，支持以 git worktree 实现隔离工作区，让 AI 在独立分支并行作业而不污染主仓库；通过 living specs 与可复用 skills 沉淀项目知识，未来还将支持自生长扩展。目前已开源，设计工程师可直接上手体验。

makecindy/cindy

Consider it done. The open-source AI agent that works out of the box · 想到，就能做到。开源、开箱即用的 AI Agent。 https://cindy.app English · 简体中文 🌐 cindy.app  ·  ⬇️ Download Cindy is an open-source AI agent that works out of the box. She brings multiple harnesses, models and tools into one agent that finishes real work in your projects and apps. Ready from day one, yours to shape over time. Cindy runs locally on your own machine, using your real files and logged-in apps. The first supported harnesses are Claude Code and Codex — more are being added, and a native harness is in the works. Models and harnesses mix freely and can switch mid-task while your workspace, memory, skills and tools stay continuous; one task can even be planned, executed in parallel, and reviewed by agents on different harness × model combos. She can drive your browser, computer and phone, and take work from IM and schedules. This repository is the open-source client for Cindy — the desktop and mobile apps plus their shared packages, organized as a pnpm monorepo. The client is free to use, and its source code is open under Apache-2.0. Bring models your way: sign in to the official Cindy service (usage deducted transparently), authorize the Claude Code / Codex Coding Plan you already pay for and keep using it inside Cindy — no duplicate bill — connect your own API keys, or use local models. See cindy.app for service details, pricing, and downloads. Linux users: see the Ubuntu, Arch Linux and Omarchy installation guide for installation, updates, keyring setup and migration from a manual install. Yours to shape Open source means more than visible — it means changeable: Memory — correct her once and she does it right from then on, shared across harnesses. Skills — teach a way of working once and reuse it everywhere; handing them to your team is in the making. Automation — recurring work schedules itself, runs itself, reports back. MCP — wire your internal tools …[truncated]

这个开源项目为 Claude Code、Cursor、Codex 等主流 AI 编程工具提供了一套完整的性能优化系统，涵盖 skills、记忆优化、安全扫描和 MCP 配置。作为 Anthropic 黑客松获奖项目，它已在生产环境经过 10 个月验证，刚发布 v2.0.0-rc.1 版本并新增 Hermes 操作员工作流。

Impeccable 是一套面向 Claude 的开源前端设计技能库，在 Anthropic 官方 frontend-design 基础上扩展了 7 大领域参考、23 条命令与 27 条反模式规则，并配有 CLI 和浏览器扩展。它通过共享设计词汇与自动审计，帮助设计工程师让 AI 输出摆脱千篇一律的 SaaS 模板感。

NVIDIA 发布开源安全扫描工具 SkillSpector，专门检测 Claude Code、Codex CLI 等 AI agent skills 中的漏洞与恶意代码。研究显示超过 26% 的 skills 存在安全漏洞。该工具支持 64 种漏洞模式识别、静态分析结合 LLM 语义评估，可在安装前完成风险评估。

Pollen Robotics 发布了 Microduck，一款仅 25 cm 高、重 800 g 的迷你双足机器鸭。它基于 Rockchip RK3566，以 50 Hz 控制循环运行强化学习策略（MuJoCo + PPO + sim2real），支持行走、滚轮模式、拾物与自主起身，并附带完整的控制代码、蓝牙手柄映射与 OTA 更新机制。

OpenBot 是一个本地优先的桌面应用，为 Codex、Claude、Grok 等 AI Agent 提供持久化独立工作空间，支持任务队列、文件共享、浏览器控制和 Agent 间协作。它把分散的 CLI 工具整合进统一的 GUI，让多模型并行和上下文管理变得可视化，对需要同时调用多个 AI 的设计工程师来说是新的工作流选项。

Agent Reach 是一款开源 CLI 工具，通过一句话安装即可让 Claude Code、Cursor 等命令行 Agent 免费阅读 Twitter、Reddit、YouTube、小红书等全网内容，无需付费 API 或复杂配置，自动解决反爬、Cookie 和 IP 封锁问题。

VoiceStudio 是 ElevenLabs 的全本地开源替代品，整合 16 个 TTS 与 11 个 ASR 引擎，支持 646 种语言的语音克隆、视频配音、转录及有声书生成。无需 API 密钥与订阅，所有数据留在本地，适合设计师直接用于交互原型配音与多语言内容创作。

Postiz 是一款近期登上 GitHub Trending 的开源社交媒体管理平台，支持 AI 代理自动编排内容、定时发布，并提供 N8N、Make.com 及公开 API 等自动化集成。对于需要批量运营设计作品或个人品牌社交媒体账号的创作者，它提供了一套可自托管的灵活方案。

Huashu Design 3.0 是支持 Claude Code、Cursor 等 agent 的 HTML-native 设计 skill。一句话即可生成网页、App 原型、PPT 及动画等可交付物，内置 84 种风格库与「设计体检」机制。项目已改为 MIT 完全开源，设计师可通过 npx 直接安装，把自然语言直接变成可上线的设计。

multica-ai/andrej-karpathy-skills

A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls. Karpathy-Inspired Claude Code Guidelines Check out my new project Multica — an open-source platform for running and managing coding agents with reusable skills. Follow me on X: https://x.com/jiayuan_jy A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls. English | 简体中文 The Problems From Andrej's post: "The models make wrong assumptions on your behalf and just run along with them without checking. They don't manage their confusion, don't seek clarifications, don't surface inconsistencies, don't present tradeoffs, don't push back when they should." "They really like to overcomplicate code and APIs, bloat abstractions, don't clean up dead code... implement a bloated construction over 1000 lines when 100 would do." "They still sometimes change/remove comments and code they don't sufficiently understand as side effects, even if orthogonal to the task." The Solution Four principles in one file that directly address these issues: Principle Addresses Think Before Coding Wrong assumptions, hidden confusion, missing tradeoffs Simplicity First Overcomplication, bloated abstractions Surgical Changes Orthogonal edits, touching code you shouldn't Goal-Driven Execution Leverage through tests-first, verifiable success criteria The Four Principles in Detail 1. Think Before Coding Don't assume. Don't hide confusion. Surface tradeoffs. LLMs often pick an interpretation silently and run with it. This principle forces explicit reasoning: State assumptions explicitly — If uncertain, ask rather than guess Present multiple interpretations — Don't pick silently when ambiguity exists Push back when warranted — If a simpler a…[truncated]