The Agent Skills Directory

[SAFE]: The analyzed files consist of technical documentation, architectural patterns, and code snippets for embedded database management. The content is educational and aligns with standard security and performance best practices.
[INDIRECT_PROMPT_INJECTION]: The skill provides patterns for processing external data within database workflows. It explicitly identifies injection risks and provides mitigation strategies, such as the use of parameterized queries and Pydantic schema validation at application boundaries.
Ingestion points: Data ingestion is primarily handled through SQL execution layers described in the parameterized-sql-only.md and json1-and-structured-data.md references.
Boundary markers: The skill mandates the use of parameterized placeholders and schema validation to separate SQL structure from external data.
Capability inventory: The skill utilizes local file system operations for database storage and provides Python-based execution recipes for data management.
Sanitization: Sanitization is enforced via parameter binding (qmark and named styles) and data validation using standard libraries.
[EXTERNAL_DOWNLOADS]: The documentation references and recommends the use of established third-party libraries including portalocker, pydantic, sqlcipher3, and keyring. These are well-known, legitimate tools within the Python and Rust ecosystems.

using-embedded-database