media-processing
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- Prompt Injection (SAFE): No patterns detected that attempt to override system prompts or bypass safety filters. The documentation contains only technical instructions relevant to the skill's purpose.
- Data Exposure & Exfiltration (SAFE): No hardcoded credentials, sensitive file paths, or network exfiltration patterns were found. The skill operates locally on provided media files.
- Obfuscation (SAFE): No obfuscated code, encoded strings, or hidden Unicode characters were detected.
- Unverifiable Dependencies & Remote Code Execution (SAFE): The skill relies on well-known system utilities (FFmpeg and ImageMagick). The Python wrapper script uses the secure
subprocess.runmethod with a list of arguments, preventing shell injection vulnerabilities. - Privilege Escalation (SAFE): The skill does not contain any commands attempting to acquire root or administrator privileges. Installation instructions for system tools correctly suggest using standard package managers.
- Persistence Mechanisms (SAFE): No code was found that attempts to establish persistence on the system.
- Metadata Poisoning (SAFE): The skill metadata correctly reflects its functionality and contains no deceptive or malicious instructions.
- Indirect Prompt Injection (SAFE): The skill processes media files but does not interpolate untrusted data into LLM prompts. Command execution is handled via a structured API that treats filenames as literals.
- Time-Delayed / Conditional Attacks (SAFE): No logic was found that gates behavior based on environmental triggers or specific dates.
- Dynamic Execution (SAFE): No dynamic code generation, runtime compilation, or unsafe deserialization (e.g., pickle) was found in the provided scripts.
Audit Metadata