The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill contains references to external websites for documentation and model comparisons.
Malicious URL Detection: Automated scanners flagged 'www.buildmvpfast.com/articles/best-llms-2026-guide/creative-writing-ai'. However, within the context of the 'Director's Bible', this is listed as a source for creative writing advice. While the scanner flags it, it does not currently represent an active payload or exfiltration point in this context.
Other references: The skill cites several external blogs and documentation pages for AI models like Midjourney, ElevenLabs, and Google Veo. These are informational and consistent with the skill's purpose as a director's guide.
[INDIRECT_PROMPT_INJECTION]: The skill defines a large attack surface for indirect prompt injection.
Ingestion points: The skill ingests user briefs and scripts, specifically in Phase 0 and Phase 1 of the grilling workflow (SKILL.md).
Boundary markers: The skill instructions do not explicitly define boundary markers for user-provided scripts, although it provides rigid structural templates for its own output.
Capability inventory: The skill does not perform subprocess calls, file writes, or direct network operations. Its primary output is natural language prompts for other AI models.
Sanitization: There is no evidence of sanitization of user-provided story ideas or character names before they are interpolated into the 'Director's Bible' framework.
[NO_CODE]: The skill consists entirely of Markdown and YAML frontmatter. There are no associated scripts (.py, .js, .sh) or executable files. All instructions are aimed at guiding the AI's persona and logic rather than executing local system commands.

director