litellm-agent-platform
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
- [EXTERNAL_DOWNLOADS]: The documentation instructs users to download the platform source code via 'git clone' from a GitHub repository (LiteLLM-Labs/litellm-agent-platform). While LiteLLM is a recognized project in the AI ecosystem, users should always verify the integrity of third-party repositories before installation.
- [PROMPT_INJECTION]: The skill facilitates an indirect prompt injection attack surface. Evidence: (1) Ingestion point: The 'input' parameter in agent execution API calls (SKILL.md); (2) Boundary markers: No delimiters or specialized instructions are shown to isolate untrusted user data; (3) Capability inventory: Agents have access to sensitive tools including 'github', 'aws', and 'filesystem' (SKILL.md); (4) Sanitization: No evidence of input validation or sanitization is present in the provided examples. Malicious content in processed data (e.g., CI logs or PR comments) could potentially influence agent behavior.
Audit Metadata