The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by processing external text files that could potentially contain malicious instructions.
Ingestion points: Reads the core contract, style rules, book plan, and canon files from the project directory.
Boundary markers: No explicit delimiters or instructions to ignore instructions within external files are defined in the workflow.
Capability inventory: The skill writes markdown files for book chapters and executes the story-state command.
Sanitization: The process does not specify any validation or sanitization for the content being processed.
[COMMAND_EXECUTION]: The workflow requires the execution of a task-specific utility named story-state to manage story metadata and chronology. This command appears to be a local tool intended for use within the book-drafting environment.
[NO_CODE]: The skill contains no executable script files (.py, .js, .sh), consisting entirely of markdown-based instructions for the agent's workflow.

write-book