fal-audio
Fail
Audited by Gen Agent Trust Hub on Mar 7, 2026
Risk Level: HIGHCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION]: Both 'scripts/speech-to-text.sh' and 'scripts/text-to-speech.sh' are vulnerable to arbitrary command execution via the '--add-fal-key' argument. The scripts take the user-provided key and append it directly to a '.env' file using 'echo "FAL_KEY=$KEY_VALUE" >> .env' without any sanitization. Subsequently, the scripts execute 'source .env', which runs the contents of the file as shell code. An attacker can exploit this by providing an input containing shell metacharacters (such as newlines or command substitutions) to execute unauthorized commands on the system.
- [CREDENTIALS_UNSAFE]: The skill implements a mechanism to store the 'FAL_KEY' API credential in plain text within a '.env' file in the local filesystem, which is an insecure practice that leads to credential exposure.
- [EXTERNAL_DOWNLOADS]: The scripts utilize 'curl' to transmit data to 'https://fal.run', which is the official API endpoint for the fal.ai service. This communication is required for the skill's primary purpose of audio processing and uses the vendor's legitimate infrastructure.
Recommendations
- AI detected serious security threats
Audit Metadata