viral-video-copywriting
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the execution of a local Python script (
extract_douyin_video.py) via a shell command to process user-provided video URLs. While necessary for the skill's primary functionality, this creates a point of interaction with the host system. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes content from external sources (Douyin video metadata and transcripts), which is a classic indirect prompt injection vector where malicious instructions could be embedded in the fetched data to influence agent behavior.
- Ingestion points: External video titles, descriptions, and transcripts are extracted via the
extract_douyin_video.pyscript and analyzed in the skill's second step. - Boundary markers: The instructions lack explicit delimiters or safety instructions to separate untrusted external content from the agent's core operational logic.
- Capability inventory: The skill possesses the capability to execute shell commands to run the extraction scripts.
- Sanitization: No sanitization, filtering, or validation of the fetched external text content is specified before the data is processed by the AI model.
Audit Metadata