OPEN SOURCE OBSERVER
从项目热度走向工程判断：追踪开源工具、模型与开发者生态。
01 / CODE
| 排名 | 项目 | 语言 | 今日新增 | 总 Star |
|---|---|---|---|---|
| 01 | cloudflare/security-audit-skillA coding-agent skill for multi-phase security audits with independently verified, machine-readable findings | JavaScript | +3,162 | 16,108 |
| 02 | sponsors/trycuaScale computer-use 2.0 with open-source drivers, cross-OS fleets, and benchmarks for training, evaluation, and data generation. | HTML | +383 | 24,304 |
| 03 | addyosmani/agent-skillsProduction-grade engineering skills for AI coding agents. | JavaScript | +547 | 96,964 |
| 04 | coder/coderSecure environments for developers and their agents | Go | +406 | 15,580 |
| 05 | anthropics/claude-codeClaude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands. | TypeScript | +482 | 146,671 |
| 06 | Open-Dev-Society/OpenStockOpenStock is an open-source alternative to expensive market platforms. Track real-time prices, set personalized alerts, and explore detailed company insights — built openly, for everyone, forever free. | TypeScript | +477 | 15,969 |
| 07 | higgsfield-ai/higgsfieldFault-tolerant, highly scalable GPU orchestration, and a machine learning framework designed for training models with billions to trillions of parameters | Jupyter Notebook | +325 | 4,904 |
| 08 | docling-project/doclingGet your documents ready for gen AI | Python | +94 | 66,990 |
| 09 | cloudflare/quiche🥧 Savoury implementation of the QUIC transport protocol and HTTP/3 | Rust | +84 | 11,992 |
| 10 | sponsors/asciimooYour own search engine | Go | +430 | 5,202 |
02 / MODELS
| 排名 | 模型 | 任务 | 收藏 | 下载 |
|---|---|---|---|---|
| 01 | Z-Image-Turbo | 文本生成图片 | 2,529 | 814,680 |
| 02 | MiniMax-H3 | — | 2,013 | 148,722 |
| 03 | Qwen3.8-27B | 视觉多模态理解 | 1,486 | 425,277 |
| 04 | DeepSeek-R1 | — | 1,384 | 927,379 |
| 05 | Qwen3.8-Flash-Next | 视觉多模态理解 | 1,237 | 59,857 |
| 06 | Qwen3-Coder-480B-A35B-Instruct | 文本生成 | 1,095 | 138,510 |
| 07 | IndexTTS-2 | 语音合成 | 982 | 209,126 |
| 08 | chatglm3-6b | — | 830 | 1,590,908 |
| 09 | Qwen3.8-2.4T-A95B | 文本生成 | 771 | 1,987 |
| 10 | Qwen-7B-Chat | 文本生成 | 757 | 693,326 |
DEVELOPER VIEW
cloudflare/security-audit-skill：如果你在维护任何对外服务，这是可以直接接入智能体工作流的安全审计能力，机器可读输出省去了人工整理报告的环节。
coder/coder：自托管、面向智能体的开发环境，适合需要合规隔离的团队,避免把代码上下文交给闭源 SaaS。
trycua：做 computer-use 相关训练或评测的话，开源驱动 + 跨 OS 机群 + 基准三者合一，能省掉大量自建成本。
docling：文档预处理到 RAG 的经典痛点，稳定更新的老项目，仍值得纳入数据管线。
模型侧：Qwen3.8 系列的视觉多模态版本适合做端到端文档/界面理解；Qwen3.8-2.4T-A95B 提供了超大规模开源权重的另一选择；IndexTTS-2 可用于需要中文语音合成的产品。
WORTH A STAR
1. cloudflare/security-audit-skill（+3162⭐）——单日涨幅断层第一，且它代表的"可验证技能"范式可能比项目本身更重要。安全审计是少数对"可信度"要求极高的场景，Cloudflare 选择开源这一形态，值得研究其技能设计方式。
2. trycua（+383⭐）——computer-use 目前最缺的不是模型而是数据与评测。它同时提供驱动、机群和基准，是这一赛道少见的完整栈。
3. coder/coder（+406⭐）——当智能体开始拥有自己的执行环境，"环境即基础设施"就成了运维问题。这个项目回答的是团队规模化使用编码智能体时绕不开的隔离与治理问题。
SOURCES & METHOD
GitHub Trending（项目热度、语言、Star） · ModelScope（模型任务、收藏与下载）
抓取日期：2026-09-20。排名是当日快照，指标可能随平台更新而变化。