The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes and updates project configuration files based on context, which creates a surface where embedded instructions could influence behavior.
Ingestion points: The skill reads references/context-contract.md and other project files in the workspace.
Boundary markers: No specific boundary markers or delimiters are used to isolate content within the ingested files.
Capability inventory: The skill possesses file-write capabilities to modify project documentation like VOICE.md and STYLE.md.
Sanitization: The skill lacks explicit sanitization for ingested content, although it requires user confirmation for any rules it infers.

cw-save