The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied SRT subtitle files through a subagent specified in references/storyboard-parser.md. This creates a surface where instructions embedded within the subtitles could attempt to influence the agent's behavior during the semantic grouping or image prompt generation phases.
Ingestion points: The srtPath file is read and analyzed by the agent in Step 3.
Boundary markers: The instructions do not define clear boundary markers or provide the agent with directives to ignore instructions contained within the SRT content.
Capability inventory: The skill and its subagents have capabilities for file reading/writing and local script execution.
Sanitization: There is no evidence of sanitization or filtering of the subtitle text before it is used for semantic analysis.
[COMMAND_EXECUTION]: The workflow relies on the execution of multiple Python scripts to orchestrate the process. For instance, scripts/check_env.py uses subprocess.run to invoke environment setup scripts, and scripts/workflow_helper.py is called repeatedly for directory initialization and video merging. This orchestration is standard for complex local workflows but involves significant shell command execution.
[EXTERNAL_DOWNLOADS]: The scripts/generate-image.py script makes network requests to the RunningHub API (www.runninghub.cn) to generate content and downloads resulting image files to the local workspace using urllib.request.

whiteboard-video-workflow