modal

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). This skill instructs placing API tokens directly into shell/CLI commands and to read specific keys from a .env, e.g. export MODAL_TOKEN_* and modal secret create ... API_KEY=sk-xxx, which requires handling secret values verbatim and can lead to exfiltration if the agent outputs them.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). Modal’s required runtime workflow can expose outsider-authored free text via user inputs to web endpoints (e.g., @modal.fastapi_endpoint/@modal.asgi_app handlers that accept text/prompt from HTTP request bodies), which the function can then pass into model inference.