blog-schema
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted blog post content to extract data for schema generation. While this presents a surface for instructions embedded in blog content to influence the agent's behavior, the skill includes built-in validation checks (character limits, date comparisons, and strict schema structures) that act as sanitization. The potential impact is low as the agent is primarily generating structured data rather than executing commands based on the content.
- Ingestion points: Processes blog post text in 'Step 1: Read Content'.
- Boundary markers: None explicitly defined for the ingestion phase.
- Capability inventory: Uses 'Read', 'Write', 'Grep', and 'Glob' tools to manage files.
- Sanitization: Implements strict validation rules in 'Step 8: Validate & Warn', including character counts, date validation, and sequence checking.
Audit Metadata