Can AI agents automate PlanetScale?

PlanetScale is automatable today through the official pscale CLI. This record is the hosted PlanetScale database platform for Vitess (MySQL) and Postgres. It is not Neon, MongoDB, Redis, or Microsoft SQL Server. pscale runs on macOS, Linux, and Windows. It creates and deletes databases and branches, runs non-interactive SQL for agents via pscale sql, dumps and restores, and calls the API through pscale api. Agents should pass --format json. Service-token auth uses PLANETSCALE_SERVICE_TOKEN_ID and PLANETSCALE_SERVICE_TOKEN, or the matching flags; pscale shell and pscale connect are interactive and should be avoided under a token. The PlanetScale API at https://api.planetscale.com/v1 manages organizations, databases, and branches. It does not read or write table data. API auth is a service token as Authorization ID:TOKEN, or an OAuth access token as Authorization Bearer. The first-party hosted MCP server at https://mcp.pscale.dev/mcp/planetscale uses OAuth by default, or a service-token secret as Authorization Bearer pscale_tkn_... in CI. Tools cover organizations, databases, branches, schema, Insights, billing, and SQL. The insights-only host https://mcp.pscale.dev/mcp/planetscale-insights-only excludes planetscale_execute_read_query and planetscale_execute_write_query. The local-only pscale mcp server has been removed. Application data still uses Postgres connection strings (ports 5432 or 6432) or Vitess MySQL passwords. There is no free plan; Base starts at paid clusters. Computer-use viability is unassessed. CLI scores 8 and is the best path because it covers lifecycle, SQL, and an API passthrough. MCP scores 7 as a useful official hosted subset. API scores 7 because it is management-only.

Best path today: cli · Overall automatability: 8/10