excel-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions to read and process external data files, creating a potential vector for indirect prompt injection if the Excel files contain adversarial instructions.
- Ingestion points: The skill makes extensive use of
pd.read_excel()to load data from external files (e.g.,data.xlsx,sales.xlsx,workbook.xlsx) into the agent's memory. - Boundary markers: There are no explicit instructions or delimiters used to warn the agent that the spreadsheet content should be treated as untrusted data or that it should ignore embedded natural language commands.
- Capability inventory: The skill includes scripts capable of reading local files, writing files to the disk (
to_excel), and generating image files (savefig), providing a set of actions an attacker might try to influence. - Sanitization: The examples show raw data processing without any validation or sanitization of cell contents to prevent the interpretation of hidden instructions.
Audit Metadata