Skills
Navigate
skills/posthog/ai-plugin/signals-scout-data-warehouse/Gen Agent Trust Hub

signals-scout-data-warehouse

Pass

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill identifies external data sources as potential attack vectors and provides specific defensive instructions.
  • Ingestion points: External server error messages (latest_error) and user-configured source/schema names (SKILL.md).
  • Boundary markers: Explicitly instructs the agent to treat external text strictly as data and never as instructions, emphasizing that error messages do not authorize actions.
  • Capability inventory: Uses execute-sql for metadata lookups and report-writing tools like emit_report and edit_report (SKILL.md).
  • Sanitization: Directs the agent to truncate long messages and quote them as untrusted snippets.
  • [COMMAND_EXECUTION]: The skill uses an SQL execution tool to query system metadata.
  • Evidence: Performs automated scans against system.source_schemas and system.data_modeling_views to monitor the integrity of armed data imports.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 21, 2026, 08:12 PM
Security Audit — agent-trust-hub — signals-scout-data-warehouse