user-deep-dive
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a specialized
query-runMCP tool to execute SQL queries against a PostHog database. These queries are well-defined, parameterized with a user-provided email and time window, and focused on analyzing product usage metrics (events, page views, insights, etc.). - [DATA_EXFILTRATION]: The skill reads user data from PostHog and Vitally to generate a profile. This is the core intended functionality for a customer success tool. There are no patterns suggesting the data is sent to unauthorized third-party domains; it is processed within the agent context to provide a summary to the requester.
- [EXTERNAL_DOWNLOADS]: The skill mentions falling back to
curlonly if the primary MCP tool is unavailable. The URLs generated for session replays point to the officialus.posthog.comdomain, which is the vendor's own infrastructure.
Audit Metadata