ask-park-video
Warn
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/render-showcase.shexecutes a Python script from a directory path provided as a command-line argument, which could allow for the execution of arbitrary code. Evidence:lanshu_skill_dir="$1",renderer="$lanshu_skill_dir/scripts/render_animated_diagram.py",python3 "$renderer". - [DYNAMIC_EXECUTION]: The shell script
scripts/render-showcase.shdynamically constructs and executes a file path based on external input, representing a dynamic code execution risk. - [INDIRECT_PROMPT_INJECTION]: The skill processes project-specific data files that could contain instructions to manipulate the agent's workflow. 1. Ingestion points:
project.json,process-log.md,subtitles/source.srt, andedit.jsonas specified inSKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: The agent is authorized to use shell commands, file writing, and external tool invocation (video-shotcraft). 4. Sanitization: No sanitization or validation of input file content is described. - [EXTERNAL_DOWNLOADS]: The documentation recommends installing the
video-shotcrafttool from an unverified third-party GitHub repository (Vincentwei1021/video-shotcraft), which the skill's logic depends on for visual track generation.
Audit Metadata