doc-reader
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted content from external documents.
- Ingestion points: The
read_documenttool (referenced inSKILL.md) extracts text from numerous file formats including PDF, Word, Excel, and images (via OCR). - Boundary markers: The instructions lack explicit boundary markers or system instructions advising the agent to ignore or delimit instructions found within the processed document text.
- Capability inventory: The skill provides data extraction capabilities via
read_document. The extracted text is then returned in a JSON envelope for agent processing. - Sanitization: There is no mention of sanitization, filtering, or instruction-detection for the extracted content before it is presented to the agent context.
Audit Metadata