✨ Three Core Advantages

🔁 Embedded AI Workflow Closed Loop

Embedded development has a gap that pure software development doesn't have: writing code is just the beginning — compiling, flashing, and debugging all require human intervention at every step.

AI modifies code → You manually compile → You manually flash → You copy errors to AI → AI fixes → You compile again...

embeddedskills hands this loop over to AI:

flowchart TD
    A["💬 Requirements Discussion"] --> B["✍️ Code Generation & Optimization"]
    B --> C["🔨 Auto Compile keil build / gcc build"]
    C -->|Errors| D["🤖 AI Reads Compiler Errors"]
    D --> B
    C -->|Success| E["⚡ Auto Flash openocd flash / jlink flash / probe-rs flash"]
    E --> F["🔬 Auto Debug & Verify"]
    F -->|Anomaly| G["🤖 AI Reads Debug Info"]
    G --> B
    F -->|Normal| H["✅ Submit Final Code"]

    style A fill:#4CAF50,color:#fff
    style H fill:#4CAF50,color:#fff
    style C fill:#2196F3,color:#fff
    style E fill:#2196F3,color:#fff
    style F fill:#FF9800,color:#fff
    style D fill:#f44336,color:#fff
    style G fill:#f44336,color:#fff

🆓 Completely Free, No AI Tool Lock-in

This project is fully open source and free (MIT license). Any AI tool that supports the Skill / CLAUDE.md / Rules protocol can use it directly, including but not limited to:

• Claude Code
• OpenAI Codex CLI
• TRAE
• Cursor, Kiro, Windsurf and other tools supporting the Skill protocol

No paid subscriptions required for any additional services. Switch AI tools freely.

⚡ Simple to Use, No Project Migration Required

Works directly on your existing projects — no changes to any project files. Supports mainstream build systems and debuggers in the industry:

One-command install, AI automatically detects project type and starts working:

npx skills add https://github.com/zhinkgit/embeddedskills -g -y

Skill Overview

[!TIP] Keil / GCC and J-Link / OpenOCD / probe-rs can be freely combined orthogonally — all six combinations work out of the box.

Installation

Method 1: npx (Recommended)

# One-click install all skills
npx skills add https://github.com/zhinkgit/embeddedskills -g -y

# Install only specific skill (e.g., only openocd)
npx skills add https://github.com/zhinkgit/embeddedskills --skill openocd -g -y

# Management
npx skills ls -g        # List installed
npx skills update -g    # Update to latest
npx skills remove -g    # Remove

Method 2: Direct Clone

# Claude Code (global)
git clone https://github.com/zhinkgit/embeddedskills.git ~/.claude/skills/embeddedskills

# Codex (global)
git clone https://github.com/zhinkgit/embeddedskills.git ~/.codex/skills

# Codex (current project only)
git clone https://github.com/zhinkgit/embeddedskills.git .codex/skills

[!NOTE] → Full Installation & Usage Guide — includes screenshot demonstrations and configuration instructions.

How It Works

Three key designs enable true autonomous AI closed-loop:

{
  "status": "ok | error",
  "action": "build",
  "summary": "Build successful, 0 errors, 2 warnings",
  "details": { "warnings": ["unused variable 'x' at main.c:42"] },
  "artifacts": { "hex": ".embeddedskills/build/output.hex" },
  "next_actions": ["flash to device"]
}

Three-layer configuration, override as needed, priority from high to low:

CLI args  ──►  skill/config.json (tool paths, hardware params)
          ──►  .embeddedskills/config.json (target chip, interface, log dirs)
          ──►  .embeddedskills/state.json (last build/flash/debug record)
          ──►  Defaults

Unified Log Directories:

workspace/
└── .embeddedskills/
    ├── build/          ← Build logs and hex/bin artifacts
    └── logs/
        ├── serial/     ← Serial monitor logs
        ├── can/        ← CAN message logs
        └── net/        ← Network capture files

External Dependencies

Progress

Star History

Welcome to submit Issues and PRs. Thanks to the Linux.do community for their support.