optimize-postgres
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements production-grade security practices for database engineering, such as mandatory lock timeouts and non-privileged connection pooling. Its integration with Neon, a well-known database service, for ephemeral branching is a legitimate development workflow.
- [INDIRECT_PROMPT_INJECTION]: The skill's potential for indirect prompt injection is addressed through explicit safety guardrails.
- Ingestion points: Source code and database metadata read via
read_file,search_code, andexecute_command. - Boundary markers: None present.
- Capability inventory: Access to
execute_command,write_file, andedit_file. - Sanitization: Security guardrail ASI05 explicitly prohibits the interpolation of untrusted identifiers into DDL statements or query strings.
Audit Metadata