parallax-landing-page
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a bundled Python script (
extract_frames.py) and various shell commands (mkdir,cp,python -m http.server) to process video files and organize the project. The Python script usessubprocess.runwith list-based arguments to callffmpegandffprobe, which is a safe practice that prevents shell injection. This activity is directly tied to the skill's primary purpose of video processing. - [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided 'Themes' and 'Topics' to generate marketing copy for the landing page. While this presents an attack surface where malicious input could attempt to influence the agent's output, the skill mitigates this by using a highly structured template with 5 specific scenes and strict copy guidelines (kicker, headline, script). This structure limits the potential for unauthorized instruction execution through processed data.
Audit Metadata