The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external database metadata, query logs, and execution plans, which serves as a potential surface for indirect prompt injection.
Ingestion Points: The agent processes database schemas, migration files, and EXPLAIN plan output as detailed in SKILL.md and review-checklist-triage.md.
Boundary Markers: The skill utilizes standard instructional boundaries but lacks explicit delimiters for separating potentially untrusted database log content from system instructions.
Capability Inventory: The skill facilitates the construction and execution of SQL queries through the postgres.js, pg, and Bun.SQL client drivers.
Sanitization: The skill mandates strict sanitization protocols, including the use of parameterized queries ($1), tagged templates, and the format('%I') helper to prevent SQL injection and identifier-based attacks.

postgres