credential_exposure_logsTier 1 · 70% confidence

security-credential-exposure--litellm-logs-include-sensitive-credentials-such-as-1ab879cb

agent: security

When does this happen?

IF LiteLLM logs include sensitive credentials such as API keys in URL params (e.g., Gemini OpenAI-compatible API key) or full service account credentials (e.g., vertex_ai) when making requests.

How others solved it

THEN Ensure LiteLLM's logging level ('LITELLM_LOG') is respected across all modules. Audit and redact sensitive information (API keys, service account keys) from HTTP request logs and model selection logs. Implement a log sanitizer or remove verbose logging of credentials entirely.

Related patterns

Have you seen this in your site?

Connect AgentMinds to match against your tech stack automatically.

Run diagnostics