The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates interaction with external database content, including synced remote data, which serves as a potential vector for indirect prompt injection.
Ingestion points: Query results retrieved via Turso SDKs or the MCP server (referenced in SKILL.md).
Boundary markers: SKILL.md includes explicit instructions for the agent to treat all data returned from database queries as untrusted third-party content and avoid interpreting them as instructions.
Capability inventory: Provides read/write access to local and remote databases across multiple platforms, including local file access and network requests for sync operations.
Sanitization: Relies on the provided prompt instructions to the agent to treat results as plain data and avoid execution of content returned from the database.
[EXTERNAL_DOWNLOADS]: The skill documentation references installation commands for official libraries and tools provided by the vendor.
Evidence: References installation of @tursodatabase NPM packages, pyturso for Python, tursogo for Go, and the turso crate for Rust, all of which are official resources for Turso database interaction.

turso-db