The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted external data (video files, URLs, and screenshots) to extract parameters that are subsequently used to optimize "prompt-matrix" and "industry skills" templates. This creates an indirect prompt injection surface.
Ingestion points: Processes external video sources and URLs as defined in the 'Video Input' step of SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded content were found in the processing workflow.
Capability inventory: Performs file writing operations to the ~/.claude/case-library/ directory to store analysis results.
Sanitization: No explicit validation or sanitization of extracted video metadata is mentioned before it is archived or used for template updates.
[DATA_EXPOSURE]: The skill manages a local database of analyzed content within the hidden directory ~/.claude/case-library/. This is a persistent storage mechanism for metadata extracted by the agent.
[COMMAND_EXECUTION]: The documentation includes example bash commands (using grep) intended for the user to manually search through the archived JSON files; these commands are not automated by the skill logic.

video-analysis