Sandbox
@beriberikix/zephyr-agent-skills

Zephyr RTOS skills registry for Claude Code and zephyr-cli

This repository is a registry of Agent Skills for Zephyr RTOS development. It uses `skill-meta.yaml` files and generated `index.json` data so `zephyr-cli` can pick the right skill deterministically, and it also ships as a Claude Code plugin marketplace.

63 stars15 forksPythonUpdated 4mo ago
Who it's for

Builders who want an agent to choose the right Zephyr RTOS skill for a task instead of guessing.

What it delivers

You can install the right Zephyr RTOS skill for a task and keep agent work more consistent.

What it does

Deterministic skill matching

`zephyr-cli skills suggest` scores a task against keywords, aliases, Kconfig patterns, and devicetree compatibles, then returns ranked matches with reasons.

Skill installation into a workspace

`zephyr-cli skills install <skill>` fetches the chosen skill into your workspace for use on the current project.

Claude Code plugin marketplace

The repo can be added as a Claude Code plugin marketplace and installed as `zephyr-skills@zephyr-agent-skills`.

Generated registry files

`index.json` and `.claude-plugin/marketplace.json` are generated from the skill files and kept in sync by validation scripts.

Skill catalog by Zephyr domain

The `skills/` tree groups skills by areas like board bringup, devicetree, connectivity, testing, and kernel basics.

How to get it

  1. 1zephyr-cli is the primary way agents consume this registry. It selects the right skill…
    zephyr-cli skills suggest "enable an i2c sensor in my devicetree overlay"
    zephyr-cli skills install devicetree
  2. 2The repository can also be added as a Claude Code plugin marketplace
    claude plugin marketplace add beriberikix/zephyr-agent-skills
    claude plugin install zephyr-skills@zephyr-agent-skills
  3. 3index.json is the machine-readable registry consumed by zephyr-cli. It is a generated…
    python scripts/generate_index.py

README

Zephyr Agent Skills

A registry of professional, agent-ready skills for building with Zephyr RTOS.

This repository is a curated collection of "skills" — modular packages of knowledge, code patterns, and best practices that help AI agents and human developers build high-quality embedded systems with Zephyr. Each skill is a SKILL.md plus supporting references/, scripts/, and assets/.

Using the skills

With zephyr-cli (recommended)

zephyr-cli is the primary way agents consume this registry. It selects the right skill deterministically — no guessing:

zephyr-cli skills suggest "enable an i2c sensor in my devicetree overlay"
zephyr-cli skills install devicetree

skills suggest scores your task against every skill's keywords, aliases, Kconfig patterns, and devicetree compatibles and returns a ranked list with the reasons each matched. skills install fetches the chosen skill into your workspace.

As a Claude Code plugin

The repository can also be added as a Claude Code plugin marketplace:

claude plugin marketplace add beriberikix/zephyr-agent-skills
claude plugin install zephyr-skills@zephyr-agent-skills

The umbrella zephyr-skills skill (the repo-root SKILL.md) is a broad entry point that routes agents to zephyr-cli skills suggest for deterministic skill selection.

The registry — index.json

index.json is the machine-readable registry consumed by zephyr-cli. It is a generated artifact — do not edit it by hand. It is built from each skill's SKILL.md frontmatter plus a per-skill skill-meta.yaml sidecar:

python scripts/generate_index.py

skill-meta.yaml holds the curated matcher metadata — summary, keywords, aliases, kconfig_patterns, dts_compatible, and weight. scripts/validate_skills.py (run in CI) keeps index.json, .claude-plugin/marketplace.json, and the skill set consistent.

Skill catalog

👉 Master Skill Catalog — every skill, grouped by domain (foundations & build, hardware & peripherals, connectivity, production & advanced). Or just run zephyr-cli skills suggest and let the matcher pick.

Repository structure

.
├── skills/                       # the skills — one directory each
│   └── <skill>/
│       ├── SKILL.md              # entry point (Quick Start + Validation Checklist)
│       ├── skill-meta.yaml       # curated matcher metadata for the registry
│       ├── references/           # detailed technical guides
│       └── scripts/, assets/     # helper files (where applicable)
├── index.json                    # generated machine-readable registry (consumed by zephyr-cli)
├── SKILL.md                       # umbrella skill — Claude Code entry point
├── scripts/
│   ├── generate_index.py         # regenerates index.json
│   ├── generate_marketplace.py   # regenerates .claude-plugin/marketplace.json
│   └── validate_skills.py        # repository quality gate
└── .claude-plugin/marketplace.json

Contributing

Each skill is a self-contained directory: a SKILL.md (with ## Quick Start and ## Validation Checklist sections), a skill-meta.yaml, and optional references/, scripts/, and assets/.

To add or change a skill:

  1. Edit or create the skill under skills/<name>/.
  2. Update its skill-meta.yaml matcher metadata (keywords/aliases quality directly drives zephyr-cli skills suggest accuracy).
  3. Run python scripts/generate_index.py and python scripts/generate_marketplace.py.
  4. Run python scripts/validate_skills.py — it must pass.

Commit the regenerated index.json and marketplace.json; CI verifies they are in sync with skills/.

Files in the repo

Repository payload10 top-level entries
  • .agent
  • .claude-plugin
  • .github
  • scripts
  • skills
  • .gitignore
  • index.json
  • LICENSE
  • README.md
  • SKILL.md

Discussion (0)

Ask about usage, or say what you built with it

Sign in to join the discussion.

No comments yet. Be the first to say what this is good for.

More skills

obra/
superpowers

An agentic skills framework & software development methodology that works.

285k
1 add

Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.

117k
1 add
Vincentwei1021/
anything2explainer

Topic in, narrated explainer video out. A Claude Code / Codex skill that turns any topic into a black-canvas motion-graphics explainer video with TTS voiceover, subtitles and a chapter progress bar. Chinese or English; every frame drawn in code with Remotion.

666

Open-source AI job search: scan job portals, evaluate listings into a structured A-H report with a global 1-5 score, tailor your CV, track applications — runs locally in your AI coding CLI (Claude Code, Codex, OpenCode, Antigravity…)

71k