Proof of Concept

Plaky - origin-unchecked postMessage → persistent localStorage poisoning → open redirect

Target: app.plaky.com · Program: CAKE.com · Researcher: Suruur

file:// detected. This page must be served over HTTP(S) — its hostname is used as the redirect target. Host the poc-site folder on any static host, then open the resulting URL.

1 Attacker page — one click

The button opens https://app.plaky.com/data-receiver.html as a popup — a real top-level window on the victim's plaky origin, so the write below lands in the victim's actual localStorage (browser storage partitioning does not apply). A crafted IFRAME_PARENT_MESSAGE is then posted into it. The receiver checks only that id / slug / name keys exist — no e.origin validation.

2 Victim visits the payment flow — logged out

The victim opens https://app.plaky.com/payment/plans (e.g. from a billing email) while not signed into any workspace. Plaky's root-domain router reads USED_WORKSPACES and executes — with this page hosted at :


      

The browser lands directly on this site's own page (/hacked.html) — fully attacker-controlled. A full-fidelity Plaky login / payment-card phishing page would be served here. The poisoned entry persists until site data is cleared, so the redirect repeats on every future visit. Control: with USED_WORKSPACES cleared, the same URL shows the normal login redirect (/login?redirectUrl=%2Fpayment%2Fplans&feature=ACCESS_WS) — no attacker redirect.

Why it works: "https://app.plaky.com".replace("app", SLUG) performs a plain string replacement — a slug containing / breaks out of the hostname, and the ? absorbs the /payment/plans suffix the app appends, so the attacker controls the final host and path. All steps are client-side; no server interaction, no credentials requested, nothing sent from Plaky to the attacker host.