The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local opc-doc/ directory to generate business summaries and bottleneck hypotheses, creating a surface for potential injection.
Ingestion points: The skill reads from opc-doc/outputs/** and opc-doc/reviews/dashboard.json to gather context for the review.
Boundary markers: There are no explicit instructions or delimiters defined to ensure the agent ignores instructions potentially embedded within the data it reads.
Capability inventory: The skill utilizes a Write tool to create and append to markdown and JSON files within the workspace.
Sanitization: The instructions do not define any sanitization, validation, or escaping of the external content before it is interpolated into the agent's prompt or written back to the disk.

opc-dashboard-review