test-database-workflow
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read external test configurations, environment templates, and test commands. This creates a surface for indirect prompt injection where malicious instructions embedded in these files could attempt to override safety boundaries or misdirect database operations.
- Ingestion points: Reads test configurations, environment templates, and repository test commands (SKILL.md).
- Boundary markers: The instructions mandate confirming that the connection is an explicitly designated test target and require the agent to stop and ask if isolation is unclear (SKILL.md).
- Capability inventory: Database migration, seeding, writing, and data cleanup (SKILL.md).
- Sanitization: The skill relies on explicit verification steps to ensure targets are distinct from production/development rather than automated data sanitization.
Audit Metadata