我用真实 GitHub 需求信号，构建面向 AI Agent 工作流的小而实用的工具。
当前公开主线很简单：扫描开发者已经在问什么，把证据转成 build brief，再围绕高频需求做窄而有用的开源工具。中文用户是我优先服务和商业化的人群，英文页面保持完整，是为了连接全球开源生态和上游项目。
| 项目 | 做什么 | 为什么值得关注 |
|---|---|---|
| Agent Skill Radar | 每日扫描 AI agent skills、MCP servers、prompts、agent-native developer tools 的 GitHub 需求信号。 | 如果你想判断下一个 AI Agent 工具、内容选题或开源项目该做什么，这个仓库提供可复用的证据链。 |
最新公开证据：skill-radar report
- 跟踪 AI agents、MCP、prompt systems、context tools、developer automation 的公开 GitHub 信号。
- 按需求强度、活跃度、扩展性、内容传播价值、新颖度和饱和度评分。
- 公开报告，让判断过程可见。
- 当同类信号反复出现时，构建最小可用的 companion tool。
- 给开发者：发现值得做的 AI Agent 工具机会。
- 给创作者：把 GitHub 证据变成内容选题、图文、视频和社群讨论。
- 给小团队：用公开信号做技术选型、MCP 工具筛选和产品化验证。
- 给我自己：用公开项目建立可信 IP，再承接中文用户的咨询、训练营、模板和工具服务。
我的背景是材料与半导体封装工程。那类工作依赖证据、过程控制、失效分析和可重复决策。我把同一套方法迁移到 AI-native software：
- 观察系统；
- 测量信号；
- 定位真实阻塞；
- 做一个小修复；
- 公开迭代。
- Email: mahaochen2018@hotmail.com
- LinkedIn: haochenma-packaging
I build small, practical AI agent tools from real GitHub demand signals.
My public line is simple: scan what developers are already asking for, turn the evidence into a build brief, then ship narrow tools that make agent workflows easier to use. Chinese-speaking users are my first audience for monetization and community growth; the English version stays complete so the work can connect with the global open-source ecosystem.
| Project | What it does | Why follow |
|---|---|---|
| Agent Skill Radar | Daily GitHub radar for AI agent skills, MCP servers, prompts, and agent-native developer tools. | Follow the repo if you want evidence-backed signals for what agent tools, content angles, or open-source projects are worth building next. |
Latest public evidence: skill-radar report
- Track public GitHub signals around AI agents, MCP, prompt systems, context tools, and developer automation.
- Score opportunities by demand, freshness, extensibility, creator fit, novelty, and saturation.
- Publish the report so the reasoning is visible.
- Build the smallest useful companion tool when a signal repeats.
- Developers looking for high-signal AI agent tooling opportunities.
- Technical creators turning GitHub evidence into content and community discussions.
- Small teams evaluating MCP tools, agent workflows, and productization paths.
- Chinese-speaking builders who want practical AI-agent workflows before they become crowded.
My background is materials and semiconductor packaging engineering, where useful work depends on evidence, process control, failure analysis, and repeatable decisions. I am applying that same operating style to AI-native software:
- observe the system;
- measure the signal;
- isolate the real blocker;
- ship a small fix;
- iterate in public.
- Email: mahaochen2018@hotmail.com
- LinkedIn: haochenma-packaging