harvard-art-museums-data-engineering-pipeline

Warn

Audited by Gen Agent Trust Hub on Jun 29, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to clone a source code repository from a third-party GitHub account (Manali0711). This creates a dependency on external code that does not originate from the skill's author or a well-known service provider.\n- [COMMAND_EXECUTION]: The skill provides code for a Streamlit dashboard and an ETL pipeline that executes SQL queries and database transactions. It relies on environment variables for sensitive database credentials and performs batch data operations.\n- [PROMPT_INJECTION]: The skill ingests data from an external API (api.harvardartmuseums.org), which presents a surface for indirect prompt injection where malicious content in the museum's data could potentially influence the application's behavior.\n
  • Ingestion points: Fetches artifact records and metadata from https://api.harvardartmuseums.org/object.\n
  • Boundary markers: No explicit markers or instructions are provided to the agent to treat API data as untrusted or to ignore instructions embedded within the metadata strings.\n
  • Capability inventory: Includes database writes via mysql.connector and interactive UI rendering through the streamlit framework.\n
  • Sanitization: The code uses parameterized SQL queries (%s), which is a best practice for preventing direct SQL injection, though it does not fully eliminate risks from malicious logic embedded in processed data that might be rendered in the UI.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 29, 2026, 05:56 PM
Security Audit — agent-trust-hub — harvard-art-museums-data-engineering-pipeline