restaurant-owner
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to generate restaurant plans, creating a potential surface for prompt injection attacks.\n
- Ingestion points: User-provided restaurant data entered via arguments or in response to Step 1 in
SKILL.mdandtemplates/prompt-main.md.\n - Boundary markers: Absent. The skill does not use delimiters or instructions to prevent the agent from obeying commands embedded in user input.\n
- Capability inventory: The agent is instructed to generate and save markdown files (
.md) based on the input, as seen inSKILL.md(Step 6).\n - Sanitization: Absent. There is no mention of filtering or escaping user input before it is used in prompt templates.
Audit Metadata