typesafe-ai
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes building AI-powered features that process potentially untrusted external data, such as natural language inputs and application state.
- Ingestion points: Processes user-provided natural language and application state (e.g.,
ticket.messages[0].text) as mentioned in SKILL.md. - Boundary markers: The skill suggests using named JSON fields and backticked paths to reference and isolate nested state within prompts.
- Capability inventory: Instructions cover use cases like routing, value extraction, and verification based on the ingested content.
- Sanitization: The skill focuses on design and structure but does not specify sanitization or validation logic for the ingested external data.
Audit Metadata