The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user premises and external URLs to generate screenplays and update project files, which represents an attack surface for indirect prompt injection.
Ingestion points: The skill processes user-supplied premises and external product landing page URLs as defined in the 'Inputs' section of SKILL.md.
Boundary markers: There are no delimiters or explicit instructions in SKILL.md to ignore potential commands hidden within the ingested premise text or landing page content.
Capability inventory: The skill is designed to write script files to films/<slug>/script.md and update asset-map.json metadata based on the processed inputs.
Sanitization: No sanitization, validation, or escaping of external content is required before the agent interpolates the data into project files.

film-script