The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes subscription notes which originate from an external database and could potentially contain malicious instructions intended to influence the agent's behavior.
Ingestion points: The workflow in SKILL.md (Steps 3 and 4) uses the classic_get_active_subscription and classic_get_expired_subscription tools to read subscription details and notes.
Boundary markers: The instructions do not define specific delimiters or "ignore instructions" wrappers for the subscription notes when they are interpolated into the final report.
Capability inventory: The skill has access to modification tools such as classic_update_active_subscription_note and classic_update_expired_subscription_note (identified in the Tools used section of SKILL.md).
Sanitization: There is no explicit requirement for the agent to sanitize or validate the content of the notes for potential instructions before presenting them in the report or using them for further actions.

membership-subscription-audit