The Agent Skills Directory

[SAFE]: The skill provides a high-quality methodology for database performance tuning. It follows standard industry practices and does not include any executable scripts or external dependencies that could pose a security risk.
[INDIRECT_PROMPT_INJECTION]: The skill involves a potential attack surface by instructing the agent to ingest and analyze external data such as database execution plans, logs, and system statistics. An attacker with the ability to influence these outputs could theoretically attempt to misguide the agent's analysis. This is a standard risk for database administration tasks and is noted here as a vulnerability surface rather than a direct threat. \n
Ingestion points: Database system views (e.g., pg_stat_statements, pg_stat_activity) and query execution plans (EXPLAIN ANALYZE) referenced throughout the reference files. \n
Boundary markers: None explicitly defined to separate database output from agent instructions, though the methodology emphasizes cross-referencing multiple metrics. \n
Capability inventory: The skill provides instructions for generating SQL analysis queries but does not include tools for autonomous execution, file-system writing, or network operations. \n
Sanitization: Not applicable as the skill is instructional; recommendations are formatted for expert review.

database-performance