The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill exhibits a significant vulnerability surface where untrusted data (source code) can influence actions with side effects (file writing).
Ingestion points: The skill uses Read, Grep, and Glob tools to ingest content from the local React application's codebase in SKILL.md.
Boundary markers: There are no defined delimiters or 'ignore' instructions to prevent the agent from obeying natural language commands found within the analyzed files.
Capability inventory: The skill is granted Write and Edit permissions. A successful injection via a code comment could lead to the agent overwriting sensitive files or injecting backdoors into the application.
Sanitization: No sanitization or validation logic is defined to differentiate between 'code to analyze' and 'instructions to follow.'

api-generator