zhipu-ocr
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external image files and user-provided prompts, creating a surface for indirect prompt injection attacks where text extracted from an image might be interpreted as instructions by the agent.
- Ingestion points: Processes images from local paths and an optional prompt string passed to the execution script
scripts/ocr.py. - Boundary markers: The skill does not define specific delimiters to separate the resulting OCR text from other instructions in the agent's context.
- Capability inventory: The skill reads local files and performs network operations to the Zhipu AI API endpoint.
- Sanitization: There is no evidence of sanitization or content filtering for the text identified within images before it is returned to the agent.
Audit Metadata