fal-3d
Pass
Audited by Gen Agent Trust Hub on Mar 7, 2026
Risk Level: SAFEPROMPT_INJECTIONCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The script 'generate-3d.sh' creates an attack surface for indirect prompt injection. User-supplied values for '--prompt' and '--image-url' are inserted into a JSON payload string without escaping double quotes, which can be exploited to inject additional JSON fields and alter the API request. * Ingestion points: Input arguments to 'generate-3d.sh'. * Boundary markers: None present. * Capability inventory: Network communication with 'fal.run' via curl. * Sanitization: None performed on input strings before JSON construction.
- [CREDENTIALS_UNSAFE]: The script 'generate-3d.sh' includes an '--add-fal-key' flag that writes the provided API key directly into a plaintext '.env' file. Storing secrets in plaintext on the filesystem is a risk factor for credential exposure.
- [COMMAND_EXECUTION]: The script 'generate-3d.sh' uses the 'source' command to load environment variables from '.env' files in the local directory. If an attacker manages to place or modify a '.env' file, they can achieve arbitrary command execution within the context of the script execution.
Audit Metadata