respondent-panel
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted 'artifacts' (slogans, copy, hero text) and pass them verbatim to independent synthetic respondent agents, which could allow instructions embedded in that text to influence the agents.
- Ingestion points: User-provided text or files collected in Step 1 are processed directly.
- Boundary markers: The instructions prioritize session isolation but do not specify the use of prompt delimiters or explicit warnings to the sub-agents to ignore instructions contained within the artifact.
- Capability inventory: The workflow involves spawning multiple sub-agent instances (synthetic-respondent) to process the untrusted data.
- Sanitization: There are no instructions for sanitizing or escaping the input before it is interpolated into the prompts for the respondent agents.
Audit Metadata