The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of video transcripts and supplementary documents which could contain malicious instructions.
Ingestion points: Transcript and supplementary documents read in SKILL.md (Step 1).
Boundary markers: The instructions lack specific delimiters or "ignore embedded instructions" warnings for the processed data.
Capability inventory: The skill uses scripts/validate.py for file reading and formatting checks. No network or high-privilege capabilities are exposed to this data.
Sanitization: There is no explicit sanitization or filtering of transcript content before interpolation into the description template.
[COMMAND_EXECUTION]: The skill workflow instructs the agent to execute a local validation script.
Evidence: SKILL.md (Step 6) directs the agent to run python3 scripts/validate.py <path_to_description_file> to verify output compliance.
Scope: The command is limited to a local script provided within the skill's own directory and performs benign formatting checks.

video-description