The Agent Skills Directory

[SAFE]: The skill functions as a meta-orchestrator for screenwriting projects, sequencing tasks across multiple stages. It manages project state through a dedicated story-bible.md file without requesting elevated permissions or external connectivity.- [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface by reading the story-bible.md file to resume project state. This is a design-intent feature for persistence and is evaluated as safe due to the absence of exploitable capabilities.
Ingestion points: The skill reads project state from story-bible.md in the current directory during session initialization.
Boundary markers: The skill utilizes standard Markdown headers to delimit sections within the story bible, though no specific adversarial boundary tokens are enforced.
Capability inventory: The skill coordinates narrative tasks and references other writing-focused skills; it does not contain network request utilities, subprocess execution, or file-system modification tools beyond standard state updates.
Sanitization: Project state data is processed as narrative context for the screenwriting workflow without explicit sanitization logic.

sw-workflow