The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The documentation in references/querying.md describes the use of sequelize.query and sequelize.literal, which allow the execution of raw SQL queries.
Ingestion points: Data passed to SQL execution methods like sequelize.query and sequelize.literal (e.g., in references/querying.md and references/querying.md).
Boundary markers: The skill provides explicit warnings that raw query methods "should not be used on user-generated content."
Capability inventory: The skill enables full SQL execution capabilities against a connected database via the Sequelize instance.
Sanitization: The skill provides detailed instructions on using "Replacements" and "Bind Parameters" to safely sanitize inputs within raw queries.
[CREDENTIALS_UNSAFE]: A code example in references/typescript.md includes a connection string using root:asd123 for a localhost database. This is a common dummy credential used for local development demonstration and does not pose a threat to production environments.

sequelize-patterns