queue
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill manages the delivery of arbitrary JSON payloads from publishers to subscribers. This creates a data ingestion surface where external content could potentially contain instructions aimed at subverting the agent's logic during message processing.
- Ingestion points: User-provided data payloads in the iii::durable::publish function defined in SKILL.md.
- Boundary markers: The documentation does not specify the use of delimiters or ignore instructions for the message content.
- Capability inventory: The skill has the capability to trigger any registered function (function_id) associated with a durable:subscriber trigger.
- Sanitization: The documentation does not describe any sanitization or validation processes for the message payloads before they reach the consumer functions.
Audit Metadata