The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and interpret complex user-provided VFX task requirements, which constitutes a potential vector for indirect prompt injection.
Ingestion points: User task goals, requirement clarifications, and responses to adversarial stress-testing (the /grill-me command) defined in SKILL.md.
Boundary markers: The skill does not instruct the agent to use specific delimiters or boundary markers to isolate untrusted user input from system instructions.
Capability inventory: The skill has access to powerful tools listed in its frontmatter, including Read, Write, Bash, Glob, Grep, and Agent.
Sanitization: While formal input sanitization is missing, the skill includes a 'Safety-First' mode that mandates dry runs, explicit path confirmation, and the use of a testing specialist agent for high-risk operations.

vfx-plan