elevenlabs
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill demonstrates strong security awareness regarding credential management. It provides instructions to use
os.getenv("ELEVENLABS_API_KEY")and warns users and agents against creating.envfiles or pasting keys in shared environments, preferring centrally managed configurations. - [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for potentially untrusted data through the
/generate-voiceoverworkflow. - Ingestion points: Processes
VOICEOVER-SCRIPT.mdto extract narration text for audio generation. - Boundary markers: The instructions do not specify the use of delimiters or 'ignore' directives when the agent reads the script file.
- Capability inventory: The skill possesses capabilities to write files (
save,open("wb")), execute shell commands (ffmpegfor audio conversion), and perform network operations (ElevenLabs API requests). - Sanitization: No explicit sanitization or content validation for the narration script is described. However, because this is the primary intended use of the skill, the risk is considered negligible and is dropped from LOW to SAFE severity.
- [COMMAND_EXECUTION]: The skill references the use of
ffmpegfor converting and processing audio files. These commands are standard for the domain of audio synthesis and do not involve high-risk patterns such as piping remote content to a shell.
Audit Metadata