cite-specific-section-numbers
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides templates designed to ingest external document content into the agent's context (e.g.,
[ATTACH DOCUMENT / PASTE TEXT]). This establishes an attack surface where instructions embedded within a document could attempt to override the agent's behavior. - Ingestion points: Identified in prompt templates within
SKILL.mdused for document analysis. - Boundary markers: The templates do not include robust isolation markers (like unique delimiters) to separate document content from instructions.
- Capability inventory: The skill is entirely descriptive and lacks executable scripts, shell commands, file-system access, or network operations.
- Sanitization: There are no mechanisms provided to sanitize or filter the content of the documents being analyzed.
- [NO_CODE]: The skill consists solely of Markdown documentation and prompt templates. No executable files, scripts, or runtime configurations are provided that would interact with the host system.
Audit Metadata