All jobs

Add Lighthouse CI to a SolidJS App

Yes. An agent adds a Lighthouse CI config and a workflow job to your SolidJS app that audit the built bundle on every PR with floors it measured, and opens a PR.

Run this on RushOpens the console with this prompt ready to send, in a Project that has your repository.

What the agent does

The agent reads how the app is built and served before it audits it. In an app laid out like solidjs/solid-realworld, package.json builds with rollup -c --environment production and serves locally with serve public -l 5000 -s; rollup.config.js takes src/index.js, writes ES modules into public, deletes everything in public except index.html first through rollup-plugin-delete, and runs terser only when production is set. public/index.html loads the Conduit stylesheet from demo.productionready.io, Ionicons and four Google Fonts families from CDNs, preconnects to api.realworld.io, and loads ./index.js as a module. Routing is hash-based: src/store/createRouteHandler.js reads window.location.hash.slice(2), and src/App.js matches login, register, editor, settings, article/:slug, @user and the home page, with Settings and Auth loaded lazily. .github/workflows/gh-pages.yml runs on push to main and on pull requests, installs with npm install on Node 14 through actions/setup-node@v2, runs npm run build and publishes ./public with peaceiris/actions-gh-pages@v3 on main only. The test script is the placeholder test.

The agent adds lighthouserc.js at the root, a CommonJS file since package.json sets no type. Under ci.collect it sets staticDistDir: './public', which makes Lighthouse CI serve the build itself, url entries for /, /#/login and /#/register, which render from the bundle, and numberOfRuns: 3, because the home page fetches its article list from api.realworld.io at load and a single run would swing with that API. ci.upload uses temporary-public-storage, so each run prints a report link in the job log with no account to create. The agent does not use the lighthouse:recommended preset: with a stylesheet, icons and fonts loaded from third-party hosts, the page fails several of its audits today, and a job that is red on its first run is one people disable.

Instead ci.assert.assertions names categories. The agent runs npx @lhci/cli autorun in its sandbox against its own build three times, reads the median scores for performance, accessibility, best-practices and seo, and writes categories:accessibility, categories:best-practices and categories:seo as error with a minScore just under what it measured, and categories:performance as warn, since that score depends on the remote API and the CDN assets. The measured numbers go in the pull request so a reviewer can tighten them. A short .lighthouseci entry goes in .gitignore, because the CLI writes its reports there.

The workflow gains a lighthouse job in gh-pages.yml beside deploy, triggered by the same pull_request and push events: actions/checkout, actions/setup-node on Node 20 because @lhci/cli needs a current runtime where the existing job pins 14, npm ci from the committed package-lock.json, npm run build, and npx @lhci/cli@0.14 autorun. LHCI_GITHUB_APP_TOKEN is passed from secrets when present so results post as a status check, and the job still passes without it, printing the report links. The deploy job is untouched. The agent runs the new job's commands end to end in its sandbox and keeps the output.

Start it with one API call

POST https://api.phoenix.so/o/{org}/sessions with this body:

{
  "agent": "claude",
  "input": "Add Lighthouse CI to this SolidJS app. Read package.json, package-lock.json, rollup.config.js, public/index.html, src/index.js, src/App.js, src/store/createRouteHandler.js and .github/workflows/gh-pages.yml first. Write lighthouserc.js with staticDistDir ./public, the urls /, /#/login and /#/register, three runs, temporary-public-storage upload, and category assertions rather than a preset. Build the app, run npx @lhci/cli autorun in the sandbox, and set minScore for accessibility, best-practices and seo as errors just under the median you measured, with performance as a warning; put the measured scores in the pull request. Add a lighthouse job to gh-pages.yml on Node 20 with npm ci, npm run build and npx @lhci/cli autorun, passing LHCI_GITHUB_APP_TOKEN from secrets when present, without changing the deploy job. Add .lighthouseci to .gitignore. Quote the autorun output. Open a pull request.",
  "resources": [
    {
      "url": "https://github.com/your-org/your-app"
    }
  ]
}
  1. Get an API key. In the console, open Settings, then API keys, and create a key for your org. Send it in the Authorization header as a Bearer token.
  2. Send the request. POST the body in request.json to https://api.phoenix.so/o/{org}/sessions, with your repository as the resources URL. The run starts in its own cloud sandbox.
  3. Read what came back. Follow the run on its live stream or wait for the run.completed webhook, then read the result and its trajectory.

What comes back

The pull request adds lighthouserc.js, changes .github/workflows/gh-pages.yml (the new job) and .gitignore, and adds a section to readme.md on reading the report link. Its description quotes the autorun output from the agent's run, lists the median category scores it measured and the minScore it set from each, and names the third-party assets that hold the performance score down.

A reviewer opens the report link from the job log, checks the three audited routes are the ones they care about, and decides whether performance should move from warn to error once the CDN stylesheet is self-hosted. They also check that the job runs on pull requests from the same repository and that the deploy job still publishes only from main.

Nothing in src changes. The audit measures the app as it is; the pull request only adds the measurement and the floor, and says which audits a later change could lift.

Run it on every new front-end repository

A team that wants every front end audited starts this from its own server: a handler lists its repositories and, for each one with a build script and no lighthouserc file, sends this body to POST /o/{org}/sessions with that repository in resources. A request_id that is a UUID v5 of the repository's full name keeps a retried call from starting a second run, since the same request_id and body return the run already started.

A Rush automation with a github trigger on pull_request.opened can comment with a read of the Lighthouse report once the job is in place; its runs are read-only, so they comment and do not push. Adding the job itself is the run above.

The use case this job follows, with its automation and trigger: Hand off a long task.

Why run it on Rush?

A threshold nobody measured is either ignored or red. The agent builds the app in its own sandbox, runs Lighthouse CI three times against that build, and sets the floor from the numbers it saw, keeping the full output with the pull request, so the reviewer can see the scores behind each minScore rather than trust a round figure.

A team with several front ends on Solid, React or Vue starts the same request across all of them at once. Each run reads its own build output and routes, so a hash-routed app gets hash URLs and a static site gets its pages, and each pull request names the assets that hold its scores down.

Frequently asked questions

Why staticDistDir instead of a running server?

The app is static files in public after npm run build, and Lighthouse CI serves a staticDistDir itself, so the job needs no serve process and no port. The serve dependency in package.json stays for local development.

Why audit hash URLs?

Because createRouteHandler routes on window.location.hash, /login as a path would show the home page. /#/login and /#/register are the addresses a user sees, and both render from the bundle without the remote API.

Why is performance only a warning?

The home page fetches articles from api.realworld.io on load and the page loads its stylesheet, icons and fonts from third-party hosts, so the score varies run to run and no change in src lifts it. The pull request records the measured median and suggests self-hosting the stylesheet as the first step toward making it an error.

Does it change the deploy job or the Node version it uses?

No. The new job sets up its own Node 20 for @lhci/cli; the deploy job keeps Node 14 and its npm install, so publishing to GitHub Pages behaves as before. Moving the deploy job to npm ci and a current Node is suggested in the description as a separate change.

Do I need the Lighthouse GitHub App?

No. With temporary-public-storage the report link is in the job log. Installing the app and adding LHCI_GITHUB_APP_TOKEN as a secret turns the same results into a status check on the pull request, which the job already passes through when the secret exists.

More SolidJS jobs

The same job on other stacks

Keep reading