image-to-excel
Pass
Audited by Gen Agent Trust Hub on Aug 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
SKILL.mdfile instructs the agent to execute a shell commandpython3 scripts/gen_excel.py <output_path> '<json_data>'to process extracted data. While the input is expected to be a JSON string, passing complex data through command-line arguments can lead to argument injection if the shell escaping is handled incorrectly by the agent. - [EXTERNAL_DOWNLOADS]: The
scripts/gen_excel.pyscript contains a runtime dependency check that automatically executespip install openpyxlif the library is not found. This is a common utility pattern but involves downloading and installing third-party code from a public registry (PyPI) at runtime. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. It ingests untrusted data from an image model (Category 8).
- Ingestion points: Data is extracted from user-uploaded images via the
imagetool. - Boundary markers: The prompt includes specific formatting instructions (
Format: [[...]]) but lacks strong boundary delimiters or warnings to ignore instructions embedded within the image text. - Capability inventory: The skill has the capability to write files to the filesystem (
output_path) and execute shell commands viascripts/gen_excel.py. - Sanitization: The script performs basic JSON parsing and row padding, but does not explicitly sanitize the content of the cells for potential formula injection or script-like patterns before writing to Excel.
Audit Metadata