pgcli-data-modeler
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions require the agent to execute shell commands for dependency installation (via brew, apt, or pip) and database interaction using pgcli, psql, or docker exec.
- [CREDENTIALS_UNSAFE]: The skill directs the agent to request database credentials from the user and store them in environment variables such as PGPASSWORD, exposing sensitive data to the agent's process environment.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from database schemas and table contents, providing a surface for malicious content to influence the agent. 1. Ingestion points: Result sets from SQL queries executed via pgcli or psql in SKILL.md and references/advanced_queries.md. 2. Boundary markers: Absent. 3. Capability inventory: Shell execution (pgcli in SKILL.md) and environment variable manipulation. 4. Sanitization: Absent.
Audit Metadata