Awesome Yocto AI Agent Skills

Official-doc-first skills for AI coding agents that work on Yocto Project, OpenEmbedded, and BitBake repositories.

This repo is intentionally small enough for agents to use. It does not mirror the Yocto manuals. Instead, it gives agents reliable workflows, source-selection rules, and links to primary documentation so they can inspect the right manual for the target release before changing code.

Skills

Skill Use it for
yocto-agent-suite One entrypoint that routes broad Yocto requests to the right specialized skill.
yocto-project Layers, images, SDKs, BSPs, distro configuration, package changes, kernel integration, and Yocto Project repo onboarding.
bitbake-metadata .bb, .bbappend, .bbclass, .inc, conf/*.conf, task graphs, overrides, fetchers, signatures, and BitBake debugging.
yocto-recipe-maintenance Creating, updating, auditing, and reviewing recipes/appends with cross-distro packaging checks.
yocto-image-analysis Explaining why packages are in images, image size changes, manifests, pkgdata, and what-if changes.
yocto-build-debug Parse, fetch, compile, package, QA, sstate, cache, and performance debugging.
yocto-deploy-flash Finding deploy artifacts, flashing SD/eMMC/NAND/NOR/DFU targets, and serial boot verification.
yocto-bsp-bringup Vendor BSP import, machine configs, bootloaders, kernels, DTBs, and first boot.
yocto-kernel-bsp Kernel recipes, BSP layers, machine configuration, defconfig, config fragments, device trees, kernel providers, and board bring-up workflows.
yocto-ci-release kas/manifest workflows, CI runners, shared caches, tests, SDKs, release artifacts.
yocto-ci-kas-build kas YAML, CI pipelines, build matrices, cache policy, reproducibility, and artifact publishing.
yocto-security-ota CVEs, SBOM/SPDX, firewall policy, secure boot, hardening, OTA, and rollback planning.
yocto-security-compliance CVE checks, license manifests, SPDX/SBOM, source provenance, and compliance-sensitive recipe/image changes.

Start with yocto-agent-suite when you are not sure which one to use.

Official Source Policy

Agents using these skills should prefer primary sources in this order:

  1. The checked-out workspace and its configured Yocto release or branch.
  2. Yocto Project documentation at https://docs.yoctoproject.org/.
  3. BitBake documentation at https://docs.yoctoproject.org/bitbake/.
  4. Official source trees such as https://git.yoctoproject.org/ and https://git.openembedded.org/.
  5. Layer metadata from known project layer indexes such as https://layers.openembedded.org/.
  6. GitHub mirrors only as secondary convenience mirrors, not as the authority for issues, pull requests, or release status.
  7. Third-party skill marketplaces and blogs only as examples unless they link back to official project or vendor documentation.

The current docs entrypoint observed while creating this repo was Yocto Project 6.0-tip and BitBake 2.18, with docs last updated on July 2, 2026. Treat that as a snapshot, not a promise. Always verify the release branch in the target repo before applying version-sensitive instructions. Current vulnerability, CVE, support-status, and security advisory questions must be verified online because security data changes over time.

Note: the yoctoproject/poky GitHub mirror currently says the master branch is no longer being updated and points users toward individual bitbake, openembedded-core, meta-yocto, and yocto-docs clones or bitbake-setup. Agents should check the target branch and official docs before recommending old Poky-master workflows.

Install

These instructions install the skills for Codex-style local skill loading. Other agents may use a different skills directory, but the idea is the same: copy each skill folder into the agent's skills folder.

This does not install Yocto, BitBake, compilers, Docker, WSL, board SDKs, or flashing tools. It only gives your AI agent the workflow knowledge to help with those systems.

Windows PowerShell

  1. Open PowerShell.
  2. Clone or enter this repository:
git clone <repo-url>
cd awesome-yocto-ai-agent-skills

If the repo is already cloned, just run:

cd C:\Users\prashant.divate\Downloads\Github\awesome-yocto-ai-agent-skills
  1. Create the local skills directory:
New-Item -ItemType Directory -Force "$env:USERPROFILE\.codex\skills"
  1. Copy all skill folders:
Copy-Item -Recurse -Force .\yocto-agent-suite "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-project "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\bitbake-metadata "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-recipe-maintenance "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-image-analysis "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-build-debug "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-deploy-flash "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-bsp-bringup "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-kernel-bsp "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-ci-release "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-ci-kas-build "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-security-ota "$env:USERPROFILE\.codex\skills\"
Copy-Item -Recurse -Force .\yocto-security-compliance "$env:USERPROFILE\.codex\skills\"

  1. Restart the agent or open a new chat/session so it reloads skills.

  2. Test with a prompt like:

Use $yocto-agent-suite to help me debug a Yocto build failure.

Linux or WSL

  1. Open a terminal.
  2. Clone or enter this repository:
git clone <repo-url>
cd awesome-yocto-ai-agent-skills
  1. Create the local skills directory:
mkdir -p "$HOME/.codex/skills"
  1. Copy all skill folders:
cp -R yocto-agent-suite "$HOME/.codex/skills/"
cp -R yocto-project "$HOME/.codex/skills/"
cp -R bitbake-metadata "$HOME/.codex/skills/"
cp -R yocto-recipe-maintenance "$HOME/.codex/skills/"
cp -R yocto-image-analysis "$HOME/.codex/skills/"
cp -R yocto-build-debug "$HOME/.codex/skills/"
cp -R yocto-deploy-flash "$HOME/.codex/skills/"
cp -R yocto-bsp-bringup "$HOME/.codex/skills/"
cp -R yocto-kernel-bsp "$HOME/.codex/skills/"
cp -R yocto-ci-release "$HOME/.codex/skills/"
cp -R yocto-ci-kas-build "$HOME/.codex/skills/"
cp -R yocto-security-ota "$HOME/.codex/skills/"
cp -R yocto-security-compliance "$HOME/.codex/skills/"

  1. Restart the agent or open a new chat/session so it reloads skills.

  2. Test with a prompt like:

Use $yocto-agent-suite to find the right workflow for building and flashing my board.

Updating Later

After pulling new changes:

git pull

Repeat the copy commands for your operating system. Overwriting these folders is fine if you have not edited the installed copies manually.

Agents that support direct skill loading from a repository can use the folders directly without copying.

Plugin Marketplace

A plugin marketplace is a catalog that lets an agent app install a whole plugin bundle instead of copying skill folders one by one. For example, the Yoe repo exposes a Claude Code marketplace entry that points to its bundled .claude/skills folder.

This repo keeps the canonical skills as top-level folders because that is easiest for humans to inspect and for multiple agents to consume. To generate a Codex-style plugin bundle from those folders, run:

python scripts/package_codex_plugin.py

The generated bundle appears under dist/awesome-yocto-ai-agent-skills-plugin/ with a skills/ directory, .codex-plugin/plugin.json, and a sample marketplace file. Treat dist/ as generated output.

Repository Layout

.
|-- bitbake-metadata/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   |-- references/
|   `-- scripts/
|-- yocto-agent-suite/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   `-- references/
|-- yocto-bsp-bringup/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   `-- references/
|-- yocto-build-debug/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   |-- references/
|   `-- scripts/
|-- yocto-ci-kas-build/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   |-- references/
|   `-- scripts/
|-- yocto-ci-release/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   `-- references/
|-- yocto-deploy-flash/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   |-- references/
|   `-- scripts/
|-- yocto-image-analysis/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   |-- references/
|   `-- scripts/
|-- yocto-kernel-bsp/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   |-- references/
|   `-- scripts/
|-- yocto-project/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   |-- references/
|   `-- scripts/
|-- yocto-recipe-maintenance/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   `-- references/
|-- yocto-security-compliance/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   |-- references/
|   `-- scripts/
|-- yocto-security-ota/
|   |-- SKILL.md
|   |-- agents/openai.yaml
|   `-- references/
|-- scripts/
`-- README.md

Contribution Rules

  • Keep each SKILL.md concise and procedural.
  • Put detailed topic maps in references/.
  • Link to official docs instead of copying long manual text.
  • Mark version-sensitive guidance clearly.
  • Validate changed skills with the skill validator before publishing.