The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external data in the form of trailer plans which could serve as a vector for malicious instructions.
Ingestion points: Step 1 in SKILL.md reads TRL-### trailer plans, timelines, and audio mix data.
Boundary markers: The instructions lack specific delimiters or guardrail prompts to ignore embedded instructions within the ingested media plans.
Capability inventory: The skill utilizes audio-master, scripts/render_timeline.py, delivery-qc, asset-approval, and scripts/promo_delivery.py to process the data.
Sanitization: There is no evidence of content validation or sanitization for the trailer plans before they influence script execution.
[COMMAND_EXECUTION]: The workflow relies on the execution of multiple local scripts and system-level binaries.
Evidence: The skill calls audio-master, scripts/render_timeline.py, delivery-qc, asset-approval, and scripts/promo_delivery.py as part of its core operations.

trailer-master