specialist-data-model-qa
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill interpolates untrusted user input directly into a sub-agent's system prompt without isolation.
- Ingestion points: The variable
{the user's question}inSKILL.mdis populated by external user input. - Boundary markers: There are no delimiters (such as quotes or XML tags) or explicit safety instructions used to separate user-provided text from the sub-agent's operational instructions.
- Capability inventory: The skill utilizes the
Agenttool to spawn sub-specialists and delegate complex model auditing tasks. - Sanitization: No escaping, filtering, or validation is performed on the user input before it is embedded in the prompt template.
Audit Metadata