data-querying

Pass

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill mentions using official package managers like pnpm to execute local actions (e.g., pnpm action query-agent-native-analytics). These are references to the application's internal toolchain and not suspicious remote downloads.
  • [COMMAND_EXECUTION]: The skill describes executing internal scripts and tools using pnpm. These are standard operational patterns for this agent's environment and are used for querying data from configured providers like Jira, HubSpot, and BigQuery.
  • [DATA_EXFILTRATION]: While the skill involves reading data from various sources (BigQuery, HubSpot, Gong, etc.), this is the primary and intended purpose of the 'data-querying' skill. There is no evidence of exfiltration to unauthorized or suspicious third-party domains; data is processed and presented within the agent's chat or stored in internal analysis artifacts.
  • [REMOTE_CODE_EXECUTION]: The instructions describe a pattern for 'Writing Ad-Hoc Scripts' by creating files in actions/ and running them. While this involves code execution, it is presented as a developer workflow within the app's own directory structure using provided helpers and server libraries, which is a standard feature for this type of agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 13, 2026, 04:50 AM
Security Audit — agent-trust-hub — data-querying