The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data (user profile information and CRM metadata) to dynamically construct database queries. While the skill includes instructions to confirm the matched identity with the user, it lacks explicit sanitization or boundary markers for the interpolated data points.
Ingestion points: Tool outputs from users.me and semantic-model metadata defined in SKILL.md.
Boundary markers: Absent for the interpolation of the email/name into the query filter.
Capability inventory: The skill is restricted to reading data via api_read and does not possess write or execution capabilities.
Sanitization: No specific input validation or escaping instructions are provided for the user-supplied identifiers.
[SAFE]: The skill uses vendor-specific tools and resources (revosai/api) to facilitate its primary function. These references are consistent with the skill's stated purpose of integrating with the RevOS platform.

users