A few people have asked whether Aurelia could have skills for tools like Claude Code and Codex. That is a reasonable ask. More developers are using AI coding agents inside real projects now, and framework-specific context matters.
So we created a public repository for it:
The first skill in the repo is for Aurelia 2. It is built from the same working guidance we have been using locally: the sharp edges, the migration gotchas, the preferred project shape, and the things an agent is likely to get wrong if it only learned Aurelia from a mix of old examples and generic JavaScript advice.
Why this exists
AI coding tools are only as useful as the context they have. Aurelia has been around long enough that there is a lot of older material on the web, and not all of it applies to Aurelia 2.
That is where agents can drift. They might reach for .delegate, suggest @autoinject, set
experimentalDecorators, or invent project structure that does not match how Aurelia 2 works with
Vite today.
The skill gives the agent a better starting point. It tells it to inspect the project first, follow
the conventions already in the repo, and use current Aurelia 2 patterns. It also includes reminders
for the boring things that save time later, like making sure tslib exists when importHelpers is
enabled.
This is not a replacement for the docs. It is a working guide for coding agents so they stop guessing.
What is in the Aurelia 2 skill
The current aurelia2 skill covers day-to-day Aurelia work:
- scaffolding new Aurelia 2 apps
- creating components and routes
- working with Vite projects
- using modern binding commands
- dependency injection and decorators
- testing with Vitest and Playwright
- Storybook and Tailwind setup when the project actually needs them
- SSR, prerendering, hydration, sitemap, robots, and takeover guidance
The SSR guidance lives in a separate reference file so agents only load it when the task asks for SSR, prerendering, SEO rendering, hydration, or related debugging. That keeps the main skill focused while still giving agents enough detail when the work gets deeper.
Installing it in Codex
Codex users can install the Aurelia 2 skill directly from GitHub:
python3 ~/.codex/skills/.system/skill-installer/scripts/install-skill-from-github.py \
--repo aurelia/skills \
--path skills/aurelia2
Restart Codex after installing so it picks up the new skill.
The repository also includes Codex plugin metadata, so it can be added as a marketplace:
codex plugin marketplace add aurelia/skills
Installing it in Claude Code
Claude Code users can add the repository as a plugin marketplace:
claude plugin marketplace add aurelia/skills
claude plugin install aurelia-skills@aurelia-skills
For local testing, you can also clone the repo and load it for one session:
claude --plugin-dir /path/to/skills
A shared home for future skills
The repository is called aurelia/skills on purpose. The first skill is for Aurelia 2, but the repo
does not need to stop there.
If we add skills for migrations, documentation work, plugin development, or common production patterns, they can live in the same place. The useful thing here is not the packaging format. It is having a shared, public source of truth that the community can improve.
If you have a repeatable Aurelia workflow, we would love to see it turned into a skill. It does not have to be limited to Aurelia 2. Aurelia 1 skills are welcome too, especially for teams maintaining production apps, planning migrations, or teaching agents how older Aurelia projects are structured.
Good candidates are the things you already find yourself explaining more than once: router patterns, validation setup, forms, testing, migration checks, bundler gotchas, plugin authoring, or production debugging habits. If a coding agent keeps making the same mistake, that is probably skill material.
Open a pull request on aurelia/skills with the skill folder and the context you think agents need. The goal is simple: make working with Aurelia and modern LLM tooling a nicer experience for everyone.
If you use Claude Code, Codex, or another tool that supports this style of skill, try it on an Aurelia project and tell us where it falls short. The best version of this will come from the same place Aurelia itself comes from: people using it on real apps, finding the rough parts, and helping to smooth them out.