Automad has Broken Access Control: Unauthenticated exposure of administrator bcrypt password hashes and TOTP secrets via public API endpoint
Description
Summary
A Broken Access Control vulnerability allows an unauthenticated attacker to retrieve the bcrypt password hash of every administrator account with a single POST request. The /_api/user-collection/create-first-user setup endpoint remains publicly accessible once initial configuration is complete and returns full serialized user data in the JSON response body.
Details
Affected version: - bcrypt hash exposure: >= 2.0.0-alpha.1, <= 2.0.0-beta.27 - TOTP secret exposure: only 2.0.0-beta.27
Impact
Any Automad installation reachable over HTTP is at risk no prior account, credentials, or special network position are required to exploit this vulnerability.
Potential impacts include:
- Credential hash exposure enabling offline brute-force or dictionary attacks: bcrypt password hashes for every administrator are returned in a single unauthenticated response. While hashes are not plaintext passwords, the salt embedded in the hash is not secret it is visible in the response. Administrators using common or weak passwords are at direct risk of having their plaintext password recovered. - TOTP secret exposure: The TOTP secret is included in the response starting with version 2.0.0-beta.27, the first release introducing TOTP-based two-factor authentication. If an attacker successfully recovers a plaintext password, two-factor authentication can be bypassed entirely. *Only version 2.0.0-beta.27 is affected by this specific issue.* - Information disclosure: The response discloses the absolute filesystem path to the configuration directory. While the directory structure is publicly documented, the absolute server path may expose environment-specific information.
Remediation
Update to version 2.0.0-beta.28 or later.
This issue was reported privately and fixed prior to public disclosure.
AI Insight
LLM-synthesized narrative grounded in this CVE's description and references.
Unauthenticated attacker can retrieve bcrypt password hashes (and TOTP secrets in v2.0.0-beta.27) of all Automad administrators via a publicly accessible setup endpoint.
Vulnerability
A Broken Access Control vulnerability exists in the Automad setup endpoint at /_api/user-collection/create-first-user. This endpoint remains publicly accessible after initial configuration and, when called with a POST request, returns full serialized user data including the bcrypt password hash of every administrator account. For versions 2.0.0-beta.27 the TOTP secret is also exposed. The vulnerability affects versions >= 2.0.0-alpha.1, <= 2.0.0-beta.27 [1] [2].
Exploitation
No authentication, prior account, credentials, or special network position is required. An attacker simply sends a single POST request to the /_api/user-collection/create-first-user endpoint. The server responds with JSON containing the serialized administrator data, including bcrypt password hashes (and TOTP secrets in 2.0.0-beta.27). No user interaction or race condition is needed [1] [2].
Impact
An attacker obtains bcrypt password hashes for every administrator account. While the hashes are not plaintext passwords, the embedded salt is visible in the response, enabling offline brute-force or dictionary attacks against weak or common passwords. For 2.0.0-beta.27, the TOTP secret is also disclosed, allowing bypass of two-factor authentication if the plaintext password is recovered. Additionally, the absolute filesystem path to the configuration directory is revealed, aiding further reconnaissance [1] [2].
Mitigation
Update to version 2.0.0-beta.28 or later, which fixes the issue. The fix was applied prior to public disclosure [1] [2]. No workaround is available for affected versions.
AI Insight generated on May 27, 2026. Synthesized from this CVE's description and the cited reference URLs; citations are validated against the source bundle.
Affected products
2- Range: >= 2.0.0-alpha.1, <= 2.0.0-beta.27
Patches
0No patches discovered yet.
Vulnerability mechanics
No source-code context for this CVE — mechanics is only generated when we can read the actual fix diff. Without that, the four sections (root cause, attack vector, affected code, fix) would be speculation rather than analysis.
References
2News mentions
0No linked articles in our index yet.