Listen to this Post
How CVE-2025-65107 works:
Langfuse is an open-source LLM engineering platform.
The vulnerability affects SSO provider configurations.
In versions 2.95.0 to 2.95.11 and 3.17.0 to 3.130.x,
when AUTH__CHECK setting is not explicitly defined,
the system defaults to insecure validation.
This occurs during authentication callbacks.
An attacker can craft a malicious URL.
This URL targets the SSO callback endpoint.
It manipulates parameters like state or origin.
If an authenticated user visits this URL,
via CSRF or phishing attacks,
the request executes in their session.
Due to missing AUTH__CHECK,
the system fails to verify the request.
This allows attacker credentials to link,
to the victim’s account without consent.
Resulting in account takeover.
The vulnerability is post-authentication but exploitable via CSRF.
Patches in versions 2.95.12 and 3.131.0 enforce proper checks.
Platform: Langfuse
Version: 2.95.x and 3.17.x
Vulnerability: Account takeover CSRF
Severity: Critical
Date: 11/21/2025
Prediction: Patched in 2.95.12
What Undercode Say:
bash commands:
env | grep AUTH_
export AUTH_PROVIDER_CHECK=strict
cat config/sso.yml
codes:
AUTH_PROVIDER_CHECK: null
AUTH_PROVIDER_CHECK: “strict”
if (!AUTH_PROVIDER_CHECK) { fail(); }
How Exploit:
Craft malicious SSO URL.
Phish authenticated user.
Bypass state validation.
Link attacker account.
Takeover victim session.
Protection from this CVE:
Update to patched versions.
Set AUTH__CHECK explicitly.
Use CSRF tokens.
Monitor authentication logs.
Impact:
Unauthorized account access.
LLM data exposure.
Privilege escalation risk.
Sensitive information compromise.
🎯Let’s Practice Exploiting & Learn Patching For Free:
Sources:
Reported By: nvd.nist.gov
Extra Source Hub:
Undercode
