Back to directory

embeddedskills

embeddedskills is a free, open-source collection of embedded development and debugging skills that let an AI assistant directly operate compilers, debuggers and communication buses, closing the loop from code generation to hardware verification.

embeddedskills

embeddedskills targets a gap specific to embedded development: after code is written, a person still has to compile it, flash it, and copy error output back to the AI by hand. This project hands that whole loop to the AI assistant itself, giving it skills to build with Keil MDK or CMake-based GCC projects, flash and debug through J-Link, OpenOCD or probe-rs-compatible tools, and monitor or send data over serial, CAN/CAN-FD or Ethernet, plus a combined workflow skill that detects the project type, chooses the right toolchain, and runs the full build-flash-debug sequence.

It works directly on an existing project with no migration step, supporting Keil MDK and CMake build systems, J-Link and CMSIS-DAP/DAPLink debuggers, OpenOCD and probe-rs-compatible flashing, and serial, CAN and Ethernet buses, and installs with one command into any AI tool that supports the Skill, CLAUDE.md or Rules protocol, including Claude Code, Codex CLI, TRAE, Cursor, Kiro and Windsurf.

It is written in Python and released under the MIT license.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.