GordenImage2PPTX
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill coordinates a series of local Python scripts, including chroma_key.py and compose_pptx.py, which use standard libraries to process images and generate .pptx files. These operations are performed locally on the generated assets.\n- [EXTERNAL_DOWNLOADS]: The skill documentation recommends installing reputable Python packages such as python-pptx, pillow, and numpy to ensure the execution environment is prepared for document reconstruction.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external visual data (user-provided images) to extract text, creating an inherent vulnerability surface for indirect instructions.\n
- Ingestion points: User-supplied slide images are ingested and analyzed using vision capabilities.\n
- Boundary markers: The skill uses task-specific instructions to guide extraction, but does not provide explicit delimiters for the extracted text data in the output file.\n
- Capability inventory: The skill has permissions to write files to a task-specific root directory and execute provided Python scripts.\n
- Sanitization: The skill relies on the LLM's vision model to correctly interpret and transcribe visual text as data rather than instructions.
Audit Metadata