The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-supplied themes, instructions, and existing files to generate content across multiple stages. This creates a surface for indirect prompt injection where instructions embedded in the source data might attempt to influence the behavior of the different agents in the pipeline.
Ingestion points: User inputs for theme, tone, and external files stored in the _workspace/ directory during Phase 1.
Boundary markers: The instructions do not specify explicit delimiters or "ignore embedded instructions" warnings for the agents processing the user data.
Capability inventory: The skill can create and modify files, manage a local directory structure, and generate HTML layouts.
Sanitization: Includes a dedicated editorial-reviewer agent responsible for cross-validating outputs and ensuring consistency, which provides a layer of manual or agent-driven verification.
[SAFE]: The skill operates within a controlled _workspace/ environment. No hardcoded credentials, sensitive file access, or suspicious network exfiltration patterns were detected.

visual-storytelling