Start with the change and its trust boundaries
Ask what data the change receives, where that data comes from and which user or service can act on it. Mark trust boundaries between a browser, application, database and external service.
Review the server-side permission check for every read or change to a record. Confirm that a user cannot change an identifier to reach another account’s data.
Check unsafe input and output paths
Look for input that reaches database queries, shell commands, templates, file paths or outbound requests. Use safe library functions and validate the expected type and range.
Check that errors do not reveal secrets or internal paths. Review logs to confirm they support investigation without storing passwords or tokens.
Review tests and release impact
Add tests for both permitted and denied actions. Check whether the change affects account recovery, payment, file upload or administrative workflows.
For high-risk changes, use an independent application assessment before release. A code review is one control; it does not replace testing the deployed application.
