sandbox-self-test

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The run_suite.py script utilizes subprocess.run to execute local scripts within the repository's skills/ directory. This is a core component of the test suite's functionality for verifying the toolkit.
  • [EXTERNAL_DOWNLOADS]: The skill communicates with the HubSpot API at api.hubapi.com to perform CRM data operations and validate account metadata. This is a legitimate use of the official service for this skill's purpose.
  • [DATA_EXFILTRATION]: The skill manages a dedicated environment variable, HUBSPOT_SANDBOX_ACCESS_TOKEN. It implements mandatory account-type validation in every script to prevent accidental execution against production portals.
  • [PROMPT_INJECTION]: The skill processes data retrieved from the HubSpot API.
  • Ingestion points: Responses from various HubSpot CRM and Automation endpoints are ingested by preflight.py, seed.py, run_suite.py, and teardown.py.
  • Boundary markers: No explicit delimiters are used in the script-level data processing.
  • Capability inventory: Includes subprocess.run capabilities and full CRUD access to HubSpot CRM objects (contacts, companies, lists) and automation workflows.
  • Sanitization: Data from the API is used for internal logic and reporting without specific sanitization filters.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 12:28 AM
Security Audit — agent-trust-hub — sandbox-self-test