We check every agent UI change before it reaches you. You get the final report — what changed, what was auto-fixed, what needs your call.
See everything your agent changed. Ship with confidence, not with hope.
$ curl -fsSL https://cli.frictionscan.com/install.sh | bash
> Windows installer — coming soon
Three principles
No config files, no setup calls. Your agent installs FrictionScan and maps every route, state, viewport, and feature flag from your code — automatically.
Rule-based verification across layout, contrast, accessibility, and dozens more categories. Same result every run, no token burn, no AI guessing at your UI.
When your agent makes unintentional changes, we send them back for a fix — up to three attempts. You see the full report: what changed, what was fixed, what's pending.
How it works
Your agent handles it
One command, and FrictionScan pulls context from your code and sets itself up — no configuration required.
Every agent task, before handoff
Baseline comparison catches what changed. Deterministic where it matters, advisory where it helps.
The agent retries, up to three times
Unintentional changes loop back to your agent for auto-fix before anything reaches your review.
Full report, your decision
See what changed, what was fixed, what's pending. You decide what ships.
Early Access
We're working closely with early teams to get FrictionScan right for real agent workflows. Request an invite — one of the founders will follow up personally.