Skills
Navigate
skills/reason-machines/data-skills/harvard-artifacts-data-engineering-streamlit/Gen Agent Trust Hub

harvard-artifacts-data-engineering-streamlit

Pass

Audited by Gen Agent Trust Hub on Oct 1, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of metadata from the Harvard Art Museums API, which is an external data source.\n
  • Ingestion points: fetch_artifacts and batch_fetch_artifacts functions in SKILL.md.\n
  • Boundary markers: None present in the data transformation or display code.\n
  • Capability inventory: Database writes via mysql-connector-python and network requests via requests.\n
  • Sanitization: The primary load_to_database function uses parameterized queries (%s), which provides protection against data-driven injection during the main ETL process.\n- [DYNAMIC_EXECUTION]: The build_custom_query function in the 'Advanced Features' section uses string interpolation to build SQL queries.\n
  • Evidence: Found in the 'Custom Query Builder' code snippet in SKILL.md.\n
  • Description: The function constructs the query using f-strings for the table, group_by, and filters parameters. Using string concatenation for SQL filters (e.g., f\"{k}='{v}'\") is an unsafe pattern that could lead to SQL injection if these values are sourced from untrusted user input without validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 1, 2026, 01:35 PM
Security Audit — agent-trust-hub — harvard-artifacts-data-engineering-streamlit