Explain this query.
Build the next feature.
Keep the work in the developer workflow
Schema exploration, query tuning, and database operations usually require moving between an editor, documentation, and a database console. AI coding assistants offer a place to bring those workflows together, provided they have the right tools and database-specific guidance.
What I built
I designed and built CockroachDB plugins for Claude Code, Cursor, GitHub Copilot, and OpenAI Codex. They combine integrations with a shared CockroachDB skills library. I authored the application development and local cluster skills and hardened the library for marketplace security review.
I worked with the Cloud engineering team on MCP capabilities needed for the plugin launch. The hosted CockroachDB Cloud MCP service is team-built; my work connects that service and its capabilities to developer workflows.
The ChatGPT plugin was approved in October 2026. David Joy and I co-authored the launch article, with Cloud engineering and OpenAI collaborators helping bring the integration through to release.
Beyond a prompt and a connection
The integration has to behave correctly after installation: packaged resources must resolve, local stdio servers must start with the right configuration, and hosted tools must authenticate through OAuth. It also needs clear tool behavior and guidance suited to CockroachDB.
Alongside the plugins, I wrote the CockroachDB integration for Google's MCP Toolbox for Databases and worked with Google to get it merged. I also designed and maintain langchain-cockroachdb, covering vector and hybrid search, chat history, LangGraph checkpoints, and long-term memory.
See it in use
A developer can ask a plain-English question, inspect the schema, look at a query plan, and exercise a database workflow from their coding assistant.
The homepage includes a short recording of my Codex interaction with synthetic demo data: discovery, a bounded query, EXPLAIN, and a scratch write with UUID primary keys.