format-specific-extraction
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes complex workflows for parsing and extracting text from untrusted external files, creating a surface for indirect prompt injection.
- Ingestion points: Content extraction from Office XML (DOCX/PPTX/ODT), PDF, Archives (ZIP/TAR/7z), and Email (EML/MSG/PST) as described in
SKILL.md. - Boundary markers: The documentation mentions resource-based validation (
SecurityBudget,ZipBombValidator,DepthValidator,StringGrowthValidator) to prevent denial-of-service attacks, but it does not specify prompt-level isolation or instructions for the agent to ignore embedded commands within the extracted text. - Capability inventory: The skill is primarily instructional documentation for internal extraction logic; it does not grant the agent additional network access or file-system write permissions beyond the parsing task.
- Sanitization: Technical sanitization is applied to archive structures and recursion depth to maintain system stability, but there is no documentation regarding the sanitization or escaping of the resulting text content before it is processed by the agent.
Audit Metadata