The Agent Skills Directory

[SAFE]: The skill performs static analysis of SQL queries and execution plans using local Python scripts. No malicious patterns, network exfiltration, or dangerous command execution were detected.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted SQL queries and query explain plans provided by the user. While these inputs could contain instructions designed to influence the agent's review, the skill lacks the capabilities to perform harmful actions.\n
Ingestion points: SQL query text, schema statistics, and EXPLAIN plan output processed in scripts/sql_lint.py, scripts/explain_plan_parser.py, and scripts/cardinality_estimator.py.\n
Boundary markers: None present.\n
Capability inventory: The skill only performs read-only analysis; no network operations, file-write capabilities, or subprocess spawning were identified.\n
Sanitization: Inputs are processed via standard JSON parsing, regular expressions, and the sqlglot library.

query-validation