Just a few folks have requested whether or not Aurelia may have expertise for instruments like Claude Code and Codex.
That could be a affordable ask. Extra builders are utilizing AI coding brokers inside actual initiatives now, and
framework-specific context issues.
So we created a public repository for it:
The primary ability within the repo is for Aurelia 2. It’s constructed from the identical working steering we have now
been utilizing regionally: the sharp edges, the migration gotchas, the popular challenge form, and the
issues an agent is more likely to get incorrect if it solely discovered Aurelia from a mixture of outdated examples and
generic JavaScript recommendation.
Why this exists
AI coding instruments are solely as helpful because the context they’ve. Aurelia has been round lengthy sufficient
that there’s a lot of older materials on the net, and never all of it applies to Aurelia 2.
That’s the place brokers can drift. They could attain for .delegate, recommend @autoinject, set
experimentalDecorators, or invent challenge construction that doesn’t match how Aurelia 2 works with
Vite at the moment.
The ability offers the agent a greater start line. It tells it to examine the challenge first, observe
the conventions already within the repo, and use present Aurelia 2 patterns. It additionally consists of reminders
for the boring issues that save time later, like ensuring tslib exists when importHelpers is
enabled.
This isn’t a substitute for the docs. It’s a working information for coding brokers in order that they cease
guessing.
What’s within the Aurelia 2 ability
The present aurelia2 ability covers day-to-day Aurelia work:
- scaffolding new Aurelia 2 apps
- creating parts and routes
- working with Vite initiatives
- utilizing fashionable binding instructions
- dependency injection and interior decorators
- testing with Vitest and Playwright
- Storybook and Tailwind setup when the challenge truly wants them
- SSR, prerendering, hydration, sitemap, robots, and takeover steering
The SSR steering lives in a separate reference file so brokers solely load it when the duty asks for
SSR, prerendering, website positioning rendering, hydration, or associated debugging. That retains the principle ability centered
whereas nonetheless giving brokers sufficient element when the work will get deeper.
Putting in it in Codex
Codex customers can set up the Aurelia 2 ability immediately from GitHub:
python3 ~/.codex/expertise/.system/skill-installer/scripts/install-skill-from-github.py
--repo aurelia/expertise
--path expertise/aurelia2
Restart Codex after putting in so it picks up the brand new ability.
The repository additionally consists of Codex plugin metadata, so it may be added as a market:
codex plugin market add aurelia/expertise
Putting in it in Claude Code
Claude Code customers can add the repository as a plugin market:
claude plugin market add aurelia/expertise
claude plugin set up aurelia-skills@aurelia-skills
For native testing, you too can clone the repo and cargo it for one session:
claude --plugin-dir /path/to/expertise
A shared house for future expertise
The repository is known as aurelia/expertise on function. The primary ability is for Aurelia 2, however the repo
doesn’t must cease there.
If we add expertise for migrations, documentation work, plugin growth, or frequent manufacturing
patterns, they’ll reside in the identical place. The helpful factor right here isn’t the packaging format. It’s
having a shared, public supply of fact that the group can enhance.
You probably have a repeatable Aurelia workflow, we might like to see it was a ability. It doesn’t
should be restricted to Aurelia 2. Aurelia 1 expertise are welcome too, particularly for groups sustaining
manufacturing apps, planning migrations, or educating brokers how older Aurelia initiatives are structured.
Good candidates are the stuff you already end up explaining greater than as soon as: router patterns,
validation setup, varieties, testing, migration checks, bundler gotchas, plugin authoring, or manufacturing
debugging habits. If a coding agent retains making the identical mistake, that’s most likely ability materials.
Open a pull request on
aurelia/expertise
with the ability folder and
the context you suppose brokers want. The aim is straightforward: make working with Aurelia and fashionable LLM
tooling a nicer expertise for everybody.
If you happen to use Claude Code, Codex, or one other software that helps this model of ability, strive it on an
Aurelia challenge and inform us the place it falls quick. The most effective model of this can come from the identical
place Aurelia itself comes from: folks utilizing it on actual apps, discovering the tough elements, and serving to
to easy them out.
