The Agent Skills Directory

[SAFE]: The skill provides professional instructional content for SQL performance tuning and data modeling. No malicious code, unauthorized network operations, or persistence mechanisms were detected. The behavioral traits focus on maintainability and security.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external database schema data and query results, which constitutes a potential surface for indirect prompt injection. 1. Ingestion points: Instructions mandate inspecting schema, statistics, and access paths (SKILL.md). 2. Boundary markers: No explicit delimiters are provided in the instructions. 3. Capability inventory: The skill focuses on query generation and optimization, with no specific tool permissions (like file system or network) requested in the frontmatter. 4. Sanitization: The skill explicitly directs the agent to follow SQL injection prevention and secure coding practices, which mitigates the risk of processing malicious data.

sql-pro