The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references official client libraries @libsql/client (Node.js) and libsql-client (Python) for interacting with Turso's libSQL-based database platform. These are standard packages from the service provider.
[COMMAND_EXECUTION]: The documentation includes instructions for using the turso CLI tool (e.g., turso auth login, turso db create, turso dev). These are expected administrative commands for managing the Turso platform.
[PROMPT_INJECTION]: The references/vector-search.md file contains a code example for a RAG system that is susceptible to indirect prompt injection. The example retrieves potentially untrusted content from a database and interpolates it into a prompt sent to an LLM without sufficient sanitization or boundary protection.
Ingestion points: The query function in references/vector-search.md performs a vector similarity search on the knowledge table.
Boundary markers: The prompt template uses simple labels (Context: and Question:) but lacks robust delimiters or system instructions to prevent the model from following commands embedded within the retrieved data.
Capability inventory: The provided code allows for database reads/writes and external API interaction with OpenAI.
Sanitization: No sanitization or filtering logic is present in the example to inspect or clean retrieved context before it is processed by the LLM.

turso-libsql