Your app's first user is an agent.

liluser points an autonomous synthetic-user persona at your live, staging, or preview URL. Scout signs up, onboards, tries a goal, and tells you what actually happened - with receipts.

Sign in with GitHub — 25 trial credits on us. This run costs 1. The report link is yours.

Read a real report before you try it.

Black-box by design.Real browserNo SDK or instrumentationLive / staging / previewEvidence attached

Fresh eyes, not scripted assertions.

Give liluser a URL and a goal. The persona explores the experience like a new user would, while the run records the useful details your end-to-end test never thought to ask about.

  1. 01 / POINT

    Drop in a URL.

    Live site, staging app, or preview deploy. No hooks, SDK, or instrumentation required.

  2. 02 / TRY

    Scout drives a real browser.

    It signs up, confirms email when needed, onboards, and tries the goal you set - with a real synthetic identity.

  3. 03 / REPORT

    Get the useful version of "it felt weird."

    Findings, screenshots, and a first-person retrospective — so you can inspect what happened.

A report that can point to the moment.

Every finding stays typed and local: severity, URL, step reference, screenshot, and what Scout was trying to do. Happy paths get green checks. Blocked paths stay visible.

See the self-review report

What happened
Findings and severity, in the context of the task the persona was trying to complete.
Where it happened
The page, step references, and available screenshots behind an observation.
What supports it
The recorded experience, with independently verified side effects distinguished from the agent's own account.

One run. Three ways in.

Use the surface that fits your loop. The output stays a report your team - or your coding agent - can act on.

For the loop

GitHub Action / CI

Run against every preview deploy and leave the report where the PR is already being discussed.

For the terminal

CLI

npx liluser init
npx liluser run
npx liluser audit

For the system

API

POST /v1/runs returns a run you can poll, link, and pass into the next step of your workflow.

Useful because it knows its limits.

Reports are engineering input.

They capture a persona's observed experience and evidence. Treat findings as a sharp starting point for your team, not an infallible product verdict.

Point Scout only at apps you can test.

You must own the target or have target-owner authorization. Runs create accounts, trigger emails, and interact with the app on your behalf.

Blocked automation is a finding.

CAPTCHA, KYC, bot detection, and other walls are reported as blockers. liluser does not bypass them.

Let Scout be first.

Try a run on a URL you are authorized to test. Sign in, no instrumentation, no ceremony.

Try a run