harvard-artifacts-data-engineering-pipeline

Warn

Audited by Gen Agent Trust Hub on Jun 12, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill directs users to clone a code repository from an unverified GitHub account (Manali0711). This introduces external code into the execution environment that is not maintained by the primary vendor or a trusted organization.
  • [COMMAND_EXECUTION]: The skill provides shell commands for environment setup, specifically cloning a repository and installing Python packages (streamlit, pandas, requests, mysql-connector-python, plotly, sqlalchemy) via pip.
  • [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface because it processes data from the Harvard Art Museums API. 1. Ingestion point: The fetch_artifacts function in SKILL.md retrieves metadata from the external API. 2. Boundary markers: No explicit delimiters or instructions are used to separate API-sourced data from agent instructions. 3. Capability inventory: The skill performs database write operations (cursor.execute) and displays results in a dashboard. 4. Sanitization: Basic string truncation is used during the transform step, but no content-based filtering or sanitization is implemented.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 12, 2026, 08:21 PM
Security Audit — agent-trust-hub — harvard-artifacts-data-engineering-pipeline