The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to read and analyze untrusted content from external files provided by the user during 'Critique' and 'Update' modes.
Ingestion points: The skill reads files from user-specified paths (e.g., @path/to/messaging.md) and directory locations like messaging/ or positioning/.
Boundary markers: There are no instructions to the agent to treat the ingested file content as data only, nor are there explicit delimiters used to isolate processed content from the agent's system instructions.
Capability inventory: The skill's primary capabilities involve reading files from the workspace and writing new artifacts to the messaging/ directory.
Sanitization: The skill does not specify any validation or sanitization routines for the content extracted from the external files before the agent processes them.

messaging-architecture