paseo-advisor
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user requests and conversation history to construct a 'briefing' for a sub-agent. It lacks explicit boundary markers or sanitization logic when interpolating user-supplied questions, topics, or forwarded skill names into the prompt.
- Ingestion points: User input (
question or topic), conversation context, and arguments provided during invocation inSKILL.md. - Boundary markers: Absent. The instructions do not specify using delimiters or 'ignore embedded instructions' warnings for the interpolated user content.
- Capability inventory: The skill utilizes
create_agent(via thepaseoskill) and theSkill toolto load and execute additional functionality. - Sanitization: None detected. User-supplied skill names and arguments are passed directly into the sub-agent's briefing instructions.
Audit Metadata