database-testing
Pass
Audited by Gen Agent Trust Hub on Apr 18, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No prompt injection or behavior override attempts were detected. The instructions are clearly intended to guide agent behavior for testing database integrity and performance.
- [SAFE]: No hardcoded credentials or sensitive data exposure risks were found. Placeholders used in code snippets and test fixtures are generic and non-sensitive.
- [SAFE]: No obfuscation techniques, hidden characters, or encoded URLs were identified in any of the skill's components.
- [SAFE]: The skill does not perform any unauthorized remote code execution or external downloads from untrusted sources. It relies on standard tools like 'jq' and 'node' for local validation.
- [SAFE]: The skill does not seek to escalate privileges or establish persistence. Operations are scoped to the database environments being tested.
- [SAFE]: No indirect prompt injection vulnerabilities were identified. While the skill processes external data like database schemas, this is consistent with its primary purpose, and no dangerous capabilities are exposed to untrusted input.
- Ingestion points: Database schemas and migration scripts in 'evals/database-testing.yaml'.
- Boundary markers: Present via structured YAML and code delimiters.
- Capability inventory: Database testing operations (read/write/migrate) and task coordination.
- Sanitization: Performed by underlying database drivers and testing frameworks.
Audit Metadata