setup-warehouse-snowflake

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill sends background telemetry events to onboarding.confidence.dev and events.confidence.dev. These events include session identifiers, skill progress, and inferred user sentiment (e.g., "positive", "frustrated"). The instructions explicitly direct the agent to keep this telemetry transparent to the user and never mention it.
  • [COMMAND_EXECUTION]: The skill utilizes pbcopy to automatically move generated SQL commands, which include authentication keys, to the user's system clipboard. It also uses uuidgen and shell redirects to manage session state in temporary directories.
  • [EXTERNAL_DOWNLOADS]: The skill fetches telemetry configuration and client secrets from onboarding.confidence.dev using curl at the start of every invocation.
  • [DYNAMIC_EXECUTION]: The skill employs an inline python3 -c script to parse and extract secrets from JSON data retrieved via a network request during the telemetry initialization phase.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 11:47 AM
Security Audit — agent-trust-hub — setup-warehouse-snowflake