The Agent Skills Directory

[SAFE]: The skill acts as a technical reference for LiteLLM, providing legitimate deployment and configuration instructions.\n- [EXTERNAL_DOWNLOADS]: Documents the use of Docker images and Helm charts from official sources like docker.litellm.ai and ghcr.io. These are trusted distribution points for the software described.\n- [COMMAND_EXECUTION]: Provides standard command-line examples using docker, helm, and curl for service installation and API interaction. These are user-facing examples for deployment and testing.\n- [CREDENTIALS_UNSAFE]: Includes placeholder credentials (e.g., sk-1234, sk-master-key) within configuration examples. These are standard documentation placeholders and do not represent actual secrets.\n- [INDIRECT_PROMPT_INJECTION]: Describes a system that handles untrusted LLM content. The documentation includes specific mitigations such as user-configurable redaction, masking, and logging controls to manage this surface.\n- [DATA_EXFILTRATION]: Mentions integrations with logging and observability platforms (Langfuse, Datadog, etc.). These are industry-standard monitoring tools and their use is explicitly configured by the user.

litellm