Turso MCP

Turso MCP connects an AI agent to Turso, the SQLite and libSQL database platform, so an assistant can work with an account the way a developer would from the dashboard or the CLI. The integration spans two levels: the organization control plane, where databases are listed, created, deleted and issued scoped auth tokens, and the individual database, where the agent inspects tables, reads schemas and runs SQL. Once connected, an assistant can answer which databases exist in a workspace, what columns a table holds, or what a query returns, without anyone switching to a SQL client. Read paths such as SELECT, PRAGMA and schema description are kept separate from write paths such as INSERT, UPDATE, DELETE and DDL, so a client can be configured for inspection only. Because libSQL supports vector columns, similarity search over embeddings stored in a Turso database is reachable from the same interface. The natural users are teams already running Turso for edge or embedded SQLite workloads who want an agent to help with schema exploration, ad hoc analysis, seeding test data, or provisioning a database per environment. Treat it as production database access: scope the token to the database the agent actually needs, prefer the read-only mode for exploration, and review any statement that changes data or schema before it executes.