The Agent Skills Directory

[SAFE]: The skill includes a dedicated security module, safe_query.py, which is used to construct SQL queries safely. It utilizes a Safe class wrapper and multiple validators—including regular expressions for identifiers and allow-lists for keywords—to prevent SQL injection by ensuring only sanitized data is interpolated into commands.
[SAFE]: All external tools and dependencies, such as the aurora-dsql-mcp-server and various language-specific database connectors, are referenced from official and well-known repositories associated with the awslabs organization.
[SAFE]: The dsql_lint tool provides deterministic validation for SQL statements, catching DSQL-specific constraints such as the requirement for asynchronous index creation and providing guidance on unfixable errors, which helps maintain the integrity of the database schema.
[SAFE]: Access control documentation correctly directs users to create purpose-specific database roles mapped to IAM identities using dsql:DbConnect, while reserving administrative privileges for initial setup only.
[INDIRECT_PROMPT_INJECTION]: The skill provides tools for executing queries and reading external documentation which represent an ingestion surface for untrusted data. This risk is mitigated through the mandatory use of the safe_query builder for all database operations and the dsql_lint tool for validating externally-sourced SQL content.

dsql