single-sheet-reading-and-analysis

Pass

Audited by Gen Agent Trust Hub on Jul 3, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to ingest and process untrusted data from external Excel files.
  • Ingestion points: The load_and_clean_data function in SKILL.md (Step 2) uses pd.read_excel to read data from a file path provided to the agent.
  • Boundary markers: There are no explicit delimiters or system instructions defined to prevent the AI from interpreting instructions embedded within the Excel spreadsheet cells.
  • Capability inventory: The skill has the capability to write files to the local environment via plt.savefig (Step 5) and df.to_parquet (Step 6). It also renders HTML in the agent's output via display(HTML) (Step 6).
  • Sanitization: While the skill performs basic data cleaning (regex numerical extraction and whitespace stripping), it does not sanitize against potential malicious text instructions within the data rows.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 3, 2026, 12:24 PM
Security Audit — agent-trust-hub — single-sheet-reading-and-analysis