cleanup-forms
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a Python script that executes locally to interact with the HubSpot API. It utilizes standard libraries such as
requestsfor network operations andpython-dotenvfor loading environment variables. - [EXTERNAL_DOWNLOADS]: The skill communicates with
https://api.hubapi.com. This is the official API endpoint for HubSpot, a well-known service provider, and is used here for its intended purpose of form management. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data (form names, submission counts, and creation dates) fetched from the HubSpot API, which could theoretically contain malicious instructions.
- Ingestion points: Form metadata is retrieved from the HubSpot Forms API in the Stage 2 inventory script.
- Boundary markers: The instructions explicitly require the agent to present a candidate list to the user and wait for manual confirmation before proceeding to Stage 3 (execution).
- Capability inventory: The skill can read form data and delete forms via API calls.
- Sanitization: The skill operates on structured data from a trusted API provider and incorporates a human-in-the-loop requirement for destructive actions.
Audit Metadata