The Agent Skills Directory

[SAFE]: The skill provides detailed and high-quality implementation patterns for major relational database engines (PostgreSQL, MySQL, SQLite) and ORMs (SQLAlchemy, Prisma, Drizzle, SQLx, GORM). It includes security-focused best practices such as mandatory parameterized queries and Argon2/bcrypt password hashing.
[EXTERNAL_DOWNLOADS]: Documentation within the skill references official installation scripts for well-known development tools and database services. These are documented for setup purposes and do not represent a malicious execution vector.
Evidence: references/sqlite-guide.md and references/serverless-databases.md include the command curl -sSfL https://get.tur.so/install.sh | bash for installing the Turso CLI.
[INDIRECT_PROMPT_INJECTION]: The skill provides an interface for the agent to design and interact with database schemas and data. While this is an attack surface for indirect prompt injection if reading untrusted data, the skill includes explicit instructions to use parameterized queries and sanitization to mitigate risks.
Evidence: SKILL.md includes security best practices: "Always use parameterized queries (prevents SQL injection)" and "Hash passwords with Argon2/bcrypt".
[COMMAND_EXECUTION]: The skill includes utility scripts for schema validation and migration generation. These scripts use standard libraries (SQLAlchemy) and operate on locally defined templates or provided connection strings as intended for developer workflows.
Evidence: scripts/validate_schema.py and scripts/generate_migration.py provide standard schema inspection and template generation functionality.

using-relational-databases