The Agent Skills Directory

[SAFE]: The skill is comprised entirely of Markdown documentation and instructional logic for managing story state. It does not include any scripts, binary files, or automated commands.
[DATA_EXPOSURE_AND_EXFILTRATION]: Access and writing operations are explicitly limited to project-local directories (kb/, work/, issues/). No instructions attempt to access sensitive system paths, environment variables, or external network resources.
[PROMPT_INJECTION]: The skill defines workflows for processing user-written chapters, which serves as an ingestion point for external data. However, the instructions focus on factual extraction and consistency checking rather than executing instructions within that content. There are no high-privilege tools or dangerous capabilities exposed to this data flow.

story-memory