The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external files and user input to generate structured reports, creating a surface for indirect prompt injection.
Ingestion points: Reads content from opc-doc/inputs/founder-profile.md, opc-doc/inputs/raw-ideas.md, and interactive user chat messages.
Boundary markers: The instructions lack explicit boundary delimiters (e.g., XML tags or clear separators) or 'ignore embedded instructions' warnings for the data being read from the input files.
Capability inventory: The skill uses file-writing capabilities to create inventory.md, scorecard.json, and current-stage.json in the opc-doc/ directory.
Sanitization: There are no instructions for sanitizing or validating the content of the ingested files before they are processed and written to the output directory.
[COMMAND_EXECUTION]: The skill requires the agent to execute file system operations to read and write project files within the opc-doc/ directory to maintain state across execution stages.

opc-resource-audit