The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute cat commands to read internal framework files (e.g., world-code/voice.md). These operations use hardcoded paths and are necessary for the skill's state management.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it processes content from multiple external files and user input. (1) Ingestion points: Files voice.md, climax.md, method.md, creation.md, and crown.md within the world-code/ directory. (2) Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present when processing file content. (3) Capability inventory: Filesystem access via Bash (cat and writing files). (4) Sanitization: Content read from the filesystem is used as context for drafting without explicit validation or escaping.

world-conversation