The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted Ruby source code and migration files, creating a surface for indirect prompt injection attacks.
Ingestion points: Files located in the engine's lib/, app/, db/migrate/, and spec/dummy/ directories.
Boundary markers: The skill instructions do not define boundary markers or explicit safety instructions to prevent the agent from following directions embedded in the code it reviews.
Capability inventory: The skill employs search tools like grep and directory checks via test to perform its analysis.
Sanitization: There is no implementation of content sanitization or filtering for the external code data processed by the agent.

review-engine