record-engage-activity
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill leverages the
chrome-cdpCLI to automate interactions within the user's browser, such as navigating to the activity portal, selecting categories, and filling form text. These operations are executed within the user's local session environment. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from the Engage platform's web content and URL-based share links to pre-fill activity forms. While this creates a surface for untrusted data ingestion, the skill mitigates risk through a mandatory review phase.
- Ingestion points: Activity history text from the Engage portal and parameters extracted from share links.
- Boundary markers: Relies on a manual review step rather than programmatic delimiters.
- Capability inventory: Includes browser navigation and interaction via
chrome-cdp, access to Outlook via SSO for data derivation, and reading local skill configuration files. - Sanitization: The instructions explicitly require the agent to present all extracted and filled data to the user for validation via a confirmation prompt before proceeding with submission.
Audit Metadata