// localhost app testing
Test the app on your machine. No deploy.
Most hosted tools need a URL their cloud can reach, so you deploy first and test second. AegisRunner runs the browser on your own machine and points its AI straight at localhost — no deploy, no staging URL, no tunnel. A real test suite comes back before you push.
$ npm install -g @aegisrunner/cli
$ export AEGIS_TOKEN=aegis_xxxxxxxx
# the browser runs on your machine — scans localhost directly
$ aegis scan-runner &
$ aegis scan --local --url http://localhost:3000 --watch
# behind a login? creds stay local — set them on the runner:
$ AEGIS_USERNAME=you@example.com AEGIS_PASSWORD=… aegis scan-runnerthe browser runs on your machine. nothing leaves your network.
// the flow
Local app → tested on your machine — no URL to hand out.
Run your app locally
Your dev server on any port — Next, Vite, Rails, Django, whatever. Nothing to deploy.
aegis scan --local
A real browser runs on your own machine and points a scan straight at localhost — directly, no tunnel, no cloud relay.
AI explores localhost
The agent drives your local app like a user — pages, forms, flows — and signs in if you set a credential (it stays on your machine).
Tests come back
A generated suite plus accessibility, SEO and functional findings — against the code on your laptop, before you push.
// why localhost testing
Fast feedback where you actually work.
No deploy required
Test the branch you’re on, right now — not after a staging deploy finishes.
Runs on your machine
The browser drives localhost directly — no tunnel, no cloud relay. Fast even on big apps, and nothing leaves your network.
Nothing exposed
No inbound port, no public URL. Your app and credentials stay on your machine; only the findings come back.
One click in your app
The Vite / Nuxt / Next plugin drops a shield into your dev app — Test this page, or the whole site, without leaving localhost.
Signs in for you
Set a username/password (or a saved role) and the AI maps it onto your login form — any field, any app.
Beyond localhost
The same CLI reaches IP-restricted staging (allowlist) or VPN/air-gapped envs (self-hosted runner).
// faq
Frequently asked
Do I have to deploy my app first?
No — that’s the point. Start the local browser with `aegis scan-runner`, then `aegis scan --local --url http://localhost:3000` runs a full scan against your app on localhost. No staging URL, no public host, no deploy. (Prefer not to run a browser locally? `aegis scan --tunnel --port 3000` relays our cloud browser instead.)
Where does the browser run?
By default, on your own machine. A real headless browser (@aegisrunner/scan-runner — no Docker needed) drives your app at localhost directly, so nothing round-trips through the cloud and big apps scan fast. The AI that decides what to test still runs in our cloud; only the browser is local.
Is my local app exposed to the internet?
No. In local mode nothing is exposed at all — no inbound port, no public URL, no tunnel. Your app’s traffic and any credentials stay on your machine; only the findings (status, generated tests) come back.
How does it test pages behind a login?
For local scans, credentials stay on your machine: set AEGIS_USERNAME / AEGIS_PASSWORD (or use a saved --role) and the local browser signs in itself — they’re never sent to the cloud. The AI maps your username onto whatever the form calls its identity field — email, username, phone.
What do I get back?
A generated end-to-end suite for the flows the AI discovered, plus accessibility (WCAG), SEO and functional findings — all keyed to your app’s real pages. Export any suite to Playwright and run it locally with `npx playwright test`.
Do I need Docker?
No. `aegis scan-runner` runs a headless browser directly via npm (the first run downloads Chromium, ~150 MB, then cached). A container image (aegisrunner1/scan-runner) is available if you prefer one.
Does this need a paid plan?
Local scans run off a project CI trigger token (Pro or Business). Once you have a token, the CLI itself is open-source and zero-dependency.
Point it at localhost in two minutes
Spin up your app, grab a CI token, run one command. Watch the AI crawl your local app and hand you tests before you deploy.