Executive brief
prompts.chat, a platform for sharing AI prompts, is vulnerable to an identity confusion flaw where attackers can impersonate other users. By exploiting a mismatch in how the system checks for unique usernames, an attacker can create an account that looks identical to a target user (e.g., "Admin" vs "admin"). This allows the attacker to hijack the victim's profile page, replace their content with malicious links or misinformation, and damage the reputation of legitimate contributors.
Technical details
The vulnerability stems from a case-sensitivity mismatch between the registration/profile update logic and the database collation. While registration uses case-insensitive checks, the profile update and OAuth registration paths use Prisma's 'findUnique' method, which defaults to case-sensitive matching in PostgreSQL. This allows an attacker to bypass uniqueness constraints by creating case-variants (e.g., 'Admin' when 'admin' exists). Because the application's display surfaces resolve usernames case-insensitively using 'findFirst' without deterministic ordering, the attacker's profile can non-deterministically replace the victim's profile on canonical URLs. This leads to profile hijacking, OpenGraph spoofing, and metadata poisoning. The issue was fixed in commit 1464475 by enforcing consistent case-insensitive uniqueness checks.
Affected products
- fka prompts.chat prior to commit 1464475
Timeline
- 2026-03-24: patched: Fix committed to repository
- 2026-04-03: disclosed: CVE published