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 | +3,019 | 13,406 |
| 02 | 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 | +442 | 146,247 |
| 03 | alibaba/open-code-reviewSecure, fast, efficient, battle-tested at Alibaba's scale. Hybrid architecture code review tool: deterministic pipelines + LLM Agent, precise line-level comments, built-in multi-language ruleset (NPE, thread-safety, XSS, SQL injection), OpenAI & Anthropic compatible. | Go | +2,724 | 36,562 |
| 04 | sponsors/affaan-mThe agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond. | JavaScript | +965 | 261,956 |
| 05 | Tencent/BrowserSkillLet AI agents use your real, logged-in browser without interrupting your work. CLI + extension for browser automation across any shell-capable AI agent. | TypeScript | +1,319 | 5,222 |
| 06 | addyosmani/agent-skillsProduction-grade engineering skills for AI coding agents. | JavaScript | +677 | 96,347 |
| 07 | TencentCloud/OctopA smarter, self-hosted AI assistant — multi-user, multi-agent. | Python | +571 | 3,920 |
| 08 | Fission-AI/OpenSpecSpec-driven development (SDD) for AI coding assistants. | TypeScript | +298 | 69,296 |
| 09 | ankitects/ankiAnki is a smart spaced repetition flashcard program | Rust | +177 | 31,180 |
| 10 | anthropics/knowledge-work-pluginsOpen source repository of plugins primarily intended for knowledge workers to use in Claude Cowork | Python | +300 | 24,850 |
02 / MODELS
| 排名 | 模型 | 任务 | 收藏 | 下载 |
|---|---|---|---|---|
| 01 | Z-Image-Turbo | 文本生成图片 | 2,528 | 813,338 |
| 02 | MiniMax-H3 | — | 2,007 | 146,816 |
| 03 | Qwen3.8-27B | 视觉多模态理解 | 1,479 | 406,755 |
| 04 | DeepSeek-R1 | — | 1,384 | 927,373 |
| 05 | Qwen3.8-Flash-Next | 视觉多模态理解 | 1,230 | 58,899 |
| 06 | Qwen3-Coder-480B-A35B-Instruct | 文本生成 | 1,095 | 138,508 |
| 07 | IndexTTS-2 | 语音合成 | 982 | 208,912 |
| 08 | chatglm3-6b | — | 830 | 1,590,864 |
| 09 | Qwen3.8-2.4T-A95B | 文本生成 | 770 | 1,983 |
| 10 | Qwen-7B-Chat | 文本生成 | 757 | 693,310 |
DEVELOPER VIEW
这波项目对新范式开发者最友好的点是约定优于集成。agent-skills、OpenSpec 这类仓库提供的是一套文件格式与目录约定，你不需要改代码就能把团队规范注入 Agent，迁移成本极低。open-code-review 的价值在于免费且可私有化：兼容 OpenAI 与 Anthropic 接口，意味着你可以接自家模型，不必把代码送出去。BrowserSkill 的实际意义更大——很多企业内部系统没有 API，Agent 若不能操作用户已登录的浏览器，就只能停在 demo 阶段，这个项目补上了最后一公里。安全审计 Skill 则适合放进 CI，把机器可读的审计结果当成门禁。
WORTH A STAR
cloudflare/security-audit-skill：云厂商出品的安全审计 Skill，多阶段流程加独立验证，机器可读输出意味着可以直接进流水线，而不是停在对话框里。
alibaba/open-code-review：阿里内部大规模验证过的混合审查方案，行级评论与多语言规则集是实打实的工程资产，接口兼容主流模型厂商，接入门槛低。
Tencent/BrowserSkill：让 Agent 使用真实登录态浏览器，跨任意可执行 shell 的 Agent 通用，是当前企业内自动化最有实用价值的一块拼图。
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：2026-09-19。排名是当日快照，指标可能随平台更新而变化。