CVE-2025-0330
📋 TL;DR
This vulnerability in berriai/litellm's proxy server leaks Langfuse API keys when team settings parsing fails, exposing sensitive credentials. Attackers gaining these keys can access the Langfuse project containing all request data. Users running litellm v1.52.1 with Langfuse integration are affected.
💻 Affected Systems
- berriai/litellm
📦 What is this software?
Litellm by Litellm
⚠️ Risk & Real-World Impact
Worst Case
Full compromise of Langfuse project data including all API requests, logs, and potentially sensitive user information stored in the project.
Likely Case
Unauthorized access to Langfuse project data, enabling data theft, surveillance of API requests, and potential lateral movement.
If Mitigated
Limited exposure if Langfuse keys are rotated immediately and access logs are monitored for suspicious activity.
🎯 Exploit Status
Exploitation requires triggering an error condition in team settings parsing to leak keys.
🛠️ Fix & Mitigation
✅ Official Fix
Patch Version: v1.52.2 or later
Vendor Advisory: https://huntr.com/bounties/661b388a-44d8-4ad5-862b-4dc5b80be30a
Restart Required: No
Instructions:
1. Update litellm to v1.52.2 or later using pip: pip install --upgrade litellm==1.52.2
2. Verify the update with: pip show litellm
3. Restart any running litellm proxy services.
🔧 Temporary Workarounds
Disable Langfuse Integration
allTemporarily disable Langfuse to prevent key leakage until patching.
Remove or comment out Langfuse configuration in your litellm settings
Rotate Langfuse Keys
allImmediately rotate exposed Langfuse API keys to limit damage.
Generate new keys in Langfuse dashboard and update litellm configuration
🧯 If You Can't Patch
- Monitor logs for error messages containing Langfuse keys and implement alerting.
- Implement network segmentation to restrict litellm proxy server access.
🔍 How to Verify
Check if Vulnerable:
Check if running litellm v1.52.1 with Langfuse integration enabled.
Check Version:
pip show litellm | grep Version
Verify Fix Applied:
Confirm litellm version is v1.52.2 or later and test team settings parsing.
📡 Detection & Monitoring
Log Indicators:
- Error logs containing Langfuse API keys
- Team settings parsing failure messages
Network Indicators:
- Unexpected outbound connections to Langfuse API from unauthorized sources
SIEM Query:
source="litellm" AND ("langfuse_secret" OR "langfuse_public_key") AND level="ERROR"