llm-patterns
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides patterns for interpolating untrusted data directly into prompt templates.
- Ingestion points: The
classifyTicketPromptfunction inSKILL.mdaccepts aticketparameter representing external, untrusted data. - Boundary markers: The prompt template in
SKILL.mddirectly embeds the input via${ticket}without using delimiters (such as XML tags) or instructions to the model to ignore embedded commands. - Capability inventory: The skill demonstrates usage of the Anthropic SDK for network requests and zod for response validation; while schema validation is a best practice, the semantic logic can be subverted by malicious input.
- Sanitization: No input sanitization or escaping techniques are documented in the provided code snippets to prevent injection attacks within the interpolated string.
Audit Metadata