The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill acts as a configuration generator that interacts with the project environment, creating a surface for indirect injection if the agent processes untrusted data.
Ingestion points: The skill is triggered by Firebase-related keywords and potentially reads project files or user context to generate rules.
Boundary markers: There are no explicit instructions or delimiters provided to help the agent distinguish between data and potential malicious instructions within the files it may read.
Capability inventory: According to the skill's frontmatter, it is permitted to use Read, Write, Edit, and Bash(gcloud:*) tools.
Sanitization: The instructions do not include validation or sanitization steps to ensure that data read from the environment does not contain adversarial content designed to influence the rule generation process.

firebase-rules-generator