plain-postgres-doctor

Pass

Audited by Gen Agent Trust Hub on May 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the plain CLI tool through uv run to execute database schema audits and diagnostic checks. It includes a command to drop unknown tables which can be executed without confirmation using the --yes flag.\n- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it processes database metadata (such as table names and suggestions) as part of its diagnostic workflow.\n
  • Ingestion points: Metadata is ingested from the JSON output of diagnostic commands in SKILL.md.\n
  • Boundary markers: Boundary markers are not utilized to distinguish between data and instructions.\n
  • Capability inventory: The agent is authorized to run CLI tools and perform file system writes for migrations.\n
  • Sanitization: There is no evidence of sanitization for the input derived from the database structure.
Audit Metadata
Risk Level
SAFE
Analyzed
May 16, 2026, 06:41 AM
Security Audit — agent-trust-hub — plain-postgres-doctor