---
title: "Comment by Mihai_LeanZero on \"One Missing Parameter Cost Me Six Hours (PortSwigger Lab)\" | Hashnode"
description: "This maps onto a pattern I've hit building server-side workflows too: the confirm endpoint was only checking that the request was shaped right, not that this sp"
url: "https://hashnode.com/posts/one-missing-parameter-cost-me-six-hours-portswigger-lab/6aae394cf99fc5bd87e38e24/comment/6aaf4f267371eee4f9e22b48"
---

[Discussion](https://hashnode.com/posts/one-missing-parameter-cost-me-six-hours-portswigger-lab/6aae394cf99fc5bd87e38e24)

Comment [One Missing Parameter Cost Me Six Hours (PortSwigger Lab)](https://hashnode.com/posts/one-missing-parameter-cost-me-six-hours-portswigger-lab/6aae394cf99fc5bd87e38e24)

[Mihai\_LeanZero](https://hashnode.com/@mihaileanzero)

[Sep 20](https://hashnode.com/posts/one-missing-parameter-cost-me-six-hours-portswigger-lab/6aae394cf99fc5bd87e38e24/comment/6aaf4f267371eee4f9e22b48)

This maps onto a pattern I've hit building server-side workflows too: the confirm endpoint was only checking that the request was shaped right, not that this specific session had actually walked through the earlier steps. That's a slightly different bug than the missing per-step authorization checks you already called out, and it's worth a second layer besides those checks. A short-lived, single-use token issued when the server renders the confirmation screen and required back in the final request would have stopped this even with a stolen cookie, because replaying the right field names without ever hitting the earlier step wouldn't produce a valid token. Doesn't replace independent authorization per step, just makes skipping the flow itself harder.

[David Essien](https://hashnode.com/@davidessien)

[Sep 20](https://hashnode.com/posts/one-missing-parameter-cost-me-six-hours-portswigger-lab/6aae394cf99fc5bd87e38e24/comment/6aaf4f267371eee4f9e22b48#reply-6aaf7c32cf33e8fafc99b41e)

Hadn't thought about it in terms of a step-token specifically, but it makes sense as the piece that closes the stolen-cookie case, which my fix doesn't touch. Good addition, appreciate you separating the two clearly.
