harvard-artifacts-collection-analytics-etl
Fail
Audited by Gen Agent Trust Hub on Jun 25, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs users to clone a repository from a third-party GitHub account (
https://github.com/Manali0711/Harvard-Artifacts-Collection-Data-Engineering-Analytics-App.git) that is not verified or associated with the skill author. - [REMOTE_CODE_EXECUTION]: Following the instructions leads to the execution of downloaded code via
streamlit run app.py. This pattern allows for the execution of arbitrary code from an untrusted source. - [PROMPT_INJECTION]: The ETL pipeline ingests data from a public API, making it susceptible to indirect prompt injection if the museum data contains malicious instructions.
- Ingestion points: Artifact metadata is fetched from
https://api.harvardartmuseums.org/objectinetl_pipeline.py. - Boundary markers: The code does not use delimiters or instructions to ignore potential commands within the fetched data.
- Capability inventory: The skill uses
requests.getfor network access,mysql.connectorfor database interaction, and provides instructions for local command execution. - Sanitization: Data transformation involves basic string truncation for database compatibility but lacks filtering for prompt-altering content.
Recommendations
- AI detected serious security threats
Audit Metadata