每个 Skill 是一个 Markdown 文件,被 AI 编程助手(如 Trae、Cursor、Copilot 等)加载后,能够引导 AI 正确完成嵌入式开发任务——从环境搭建、项目创建、编译、烧录到调试,全链路覆盖。
개요
每个 Skill 是一个 Markdown 文件,被 AI 编程助手(如 Trae、Cursor、Copilot 等)加载后,能够引导 AI 正确完成嵌入式开发任务——从环境搭建、项目创建、编译、烧录到调试,全链路覆盖。 嵌入式开发环境搭建繁琐,不同芯片、不同开发板、不同工具链之间的差异巨大。我们希望通过社区共建的方式,沉淀出一套可靠的 AI 可执行知识库,让任何开发者都能**。 1. 将 SKILL.md 文件放入 AI 编程助手的 Skills 目录(如 Trae 的 ~/.trae-cn/skills/) 2. 在对话中提及相关话题,AI 会自动加载 Skill 3. 或手动 @skill-name 调用 - :创建 芯片或工具名/SKILL.md,遵循现有格式 - :提交 PR,说明你遇到的硬件问题和解决方案 - :不同芯片系列(ESP32、nRF52、GD32 等)、不同 RTOS(FreeRTOS、RT-Thread、Zephyr)、不同调试器的实战经验 - :每个 Skill 中的步骤均标注为特定硬件的参考经验,AI 必须先确认用户硬件配置 - :标注 Windows/macOS/Linux 的差异点 - :关键步骤附带预期输出的样例 - :覆盖从环境搭建到程序运行的闭环 An open-source repository collecting and sharing . Each Skill is a Markdown file that, when loaded by AI coding assistants (Trae, Cursor, Copilot, etc.), guides the AI to correctly perform embedded development tasks — from environment setup, project creation, compilation, flashing, to debugging, covering the full workflow.
README
Embedded Skills Hub / 嵌入式开发技能集
中文版
��� 这是什么?
这是一个收集和分享 嵌入式开发 AI 技能(Skill) 的开源仓库。
每个 Skill 是一个 Markdown 文件,被 AI 编程助手(如 Trae、Cursor、Copilot 等)加载后,能够引导 AI 正确完成嵌入式开发任务——从环境搭建、项目创建、编译、烧录到调试,全链路覆盖。
��� 目标
嵌入式开发环境搭建繁琐,不同芯片、不同开发板、不同工具链之间的差异巨大。我们希望通过社区共建的方式,沉淀出一套可靠的 AI 可执行知识库,让任何开发者都能用一句话开始嵌入式开发。
��� 当前 Skills
| Skill | 描述 | 适用场景 |
|---|---|---|
| stm32_dev | STM32 纯 CLI 开发指南 | STM32 ARM Cortex-M 系列芯片开发,无需 CubeMX / CubeIDE GUI |
��� 如何使用
- 将 SKILL.md 文件放入 AI 编程助手的 Skills 目录(如 Trae 的
~/.trae-cn/skills/) - 在对话中提及相关话题,AI 会自动加载 Skill
- 或手动
@skill-name调用
��� 贡献
欢迎贡献新的 Skill 或改进现有 Skill:
- 添加新 Skill:创建
芯片或工具名/SKILL.md,遵循现有格式 - 改进现有 Skill:提交 PR,说明你遇到的硬件问题和解决方案
- 特别欢迎:不同芯片系列(ESP32、nRF52、GD32 等)、不同 RTOS(FreeRTOS、RT-Thread、Zephyr)、不同调试器的实战经验
��� Skill 编写原则
- 经验 ≠ 标准 SOP:每个 Skill 中的步骤均标注为特定硬件的参考经验,AI 必须先确认用户硬件配置
- 跨平台:标注 Windows/macOS/Linux 的差异点
- 可验证:关键步骤附带预期输出的样例
- 完整链路:覆盖从环境搭建到程序运行的闭环
��� 协议
MIT License
English Version
��� What is This?
An open-source repository collecting and sharing AI Skills for embedded development.
Each Skill is a Markdown file that, when loaded by AI coding assistants (Trae, Cursor, Copilot, etc.), guides the AI to correctly perform embedded development tasks — from environment setup, project creation, compilation, flashing, to debugging, covering the full workflow.
��� Goal
Embedded development environment setup is notoriously tedious, with huge variations across different chips, boards, and toolchains. We aim to build a reliable AI-executable knowledge base through community collaboration, so that any developer can start embedded development with a single prompt.
��� Current Skills
| Skill | Description | Use Case |
|---|---|---|
| stm32_dev | STM32 Pure CLI Development Guide | STM32 ARM Cortex-M series, no CubeMX / CubeIDE GUI needed |
��� How to Use
- Copy the SKILL.md file into your AI assistant’s Skills directory (e.g.,
~/.trae-cn/skills/for Trae) - The AI auto-loads the Skill when related topics come up in conversation
- Or manually invoke with
@skill-name
��� Contributing
Contributions of new Skills or improvements to existing ones are welcome:
- Add a new Skill: Create
chip-or-tool-name/SKILL.md, following the existing format - Improve existing Skills: Submit a PR describing your hardware scenario and solution
- Especially welcome: Real-world experience with different MCU families (ESP32, nRF52, GD32, etc.), different RTOS (FreeRTOS, RT-Thread, Zephyr), and different debug probes
��� Skill Design Principles
- Experience ≠ Standard SOP: Every step in a Skill is annotated as reference experience from specific hardware. The AI must confirm the user’s hardware configuration first.
- Cross-platform: Windows/macOS/Linux differences are clearly marked
- Verifiable: Key steps include expected output samples
- End-to-end: Covers the full loop from environment setup to running code
��� License
MIT License
추천 도구
다른 키워드를 입력하거나 필터를 제거해 보세요.
설치
npx skillfish add kukucaicndy/embedded_ai_skills