ai-product
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is instructional in nature and provides templates for building AI-powered features. No malicious behavior, obfuscation, exfiltration, or persistence patterns were detected.
- [INDIRECT_PROMPT_INJECTION]: The skill addresses the vulnerability surface where untrusted user data is processed by LLMs.
- Ingestion points: The skill provides examples (e.g.,
analyzeReview,ragGenerate) that ingest strings from external sources. - Boundary markers: Patterns demonstrate the use of system and user roles to separate developer instructions from user data.
- Capability inventory: The skill provides templates for making network requests to LLM APIs but does not contain executable logic within the skill itself to perform dangerous operations.
- Sanitization: The skill contains explicit sections on input sanitization (using regex to filter prompt injection keywords) and output validation (using Zod for schema enforcement and OpenAI's moderation API for content policy checks).
Audit Metadata