Ravensworth is a shared monitoring platform. Each customer organisation has its own alert channels, named after the organisation: yours is org.204.alerts.
The dashboard opens a long-lived stream and receives alerts on it as they fire. The channel parameter takes a list, because the dashboard subscribes to several at once when you have more than one environment.
Ravensworth's public status page names its larger customers. Organisation 881 is one of them, and its channel follows the same naming.
One change to this request triggers the vulnerability. The challenge is to use your security intuition to work out what to change. Find the value the server shouldn’t trust, change it, and send it. The response tells you whether you were right.
Hint. Receive another organisation’s alerts on your stream.
“Validating the channel the caller asked for validates the subscription that gets created.”
Authorise every element, not the first one, and fail the whole request if any element is refused rather than silently dropping it. The structural fix is to authorise at the point of subscription for each channel individually, so the loop that subscribes and the loop that checks cannot get out of step. Add a test that asks for a permitted channel followed by a forbidden one and expects a refusal.
Anywhere a parameter accepts more than one value: comma separated lists, repeated query parameters, JSON arrays of ids, bulk endpoints that take a set of records, batch GraphQL operations, multi-select filters. The check is usually written for the single case and then reused. Put something you are entitled to first and something you are not entitled to after it.
One Request Away is a daily series of challenges. They’re like code review, but for HTTP requests: in each one you’re one request away from triggering a vulnerability, and your goal is to identify what you need to change.