Edit

@pgsql chat participant

The PostgreSQL extension registers an @pgsql chat participant in GitHub Copilot Chat. Use it when you want schema-aware help about the database you already use in the extension, such as object discovery, query drafting, query explanation, or next-step suggestions.

The participant works best when you already have an active connection because it can use live schema context from that database.

Important

The @pgsql chat participant is available in Visual Studio Code only. It plugs into GitHub Copilot Chat, which isn't available in Cursor or other forks. In Cursor, use Connect AI from a database node in the Connections tree to start Agent mode with the same schema context, or invoke the extension's PostgreSQL tools directly from Cursor's native chat through the MCP server.

Prerequisites

Before you start, make sure you have:

  • GitHub Copilot or GitHub Copilot Chat installed and active in Visual Studio Code.
  • The PostgreSQL extension installed.
  • Copilot/AI features enabled in the extension (pgsql.copilot.enable is true).
  • A PostgreSQL connection configured in the extension.

Note

If you ask @pgsql a question without an active connection, it can still answer general PostgreSQL questions, but it will not be able to reference your actual schema, tables, or columns.

Start a schema-aware chat

  1. Connect to the target database in the PostgreSQL extension.
  2. Open GitHub Copilot Chat.
  3. Type @pgsql followed by your question.
  4. Review the answer and decide whether you need a follow-up question, a query draft, or a deeper workflow in Agent mode.

Bring database context into Agent mode

If you already know which database you want to work with, start from the database node in the Connections tree and select Connect AI to open Agent mode with database context already in scope. This path works in both Visual Studio Code and Cursor.

Use that path when you want the AI assistant to connect, inspect schema, and work through a task such as:

  • "What tables are in this database?"
  • "Which tables look related to customers and orders?"
  • "Write a query that shows the top 10 customers by order count."

Ask questions that work well with @pgsql

The @pgsql chat participant is especially useful for these task types:

Goal Example prompt
Explore schema @pgsql What tables are in the public schema?
Understand relationships @pgsql Which tables connect orders to customers?
Draft SQL @pgsql Write a query that shows monthly revenue by region.
Explain existing SQL @pgsql Explain what this query is doing and where it might be slow.
Plan next steps @pgsql I need to add a customer status field. What should I review before I change the schema?

Ask for one decision at a time when you want the clearest results. If you need a multi-step workflow that connects, inspects schema, runs a query, and returns results in one conversation, move to Agent mode.

Review and use the response

Use @pgsql responses as working guidance, not as unreviewed final output.

  • Review generated SQL before you run it.
  • Use the query editor when you want to test or refine a query manually.
  • Use the Connections tree or the schema visualizer to confirm the objects the participant mentions.
  • Use Agent mode when you want the AI assistant to take actions instead of only answering.

Tip

Start with schema-discovery questions before you ask for large query rewrites. The better the participant understands your database context, the better the next answers tend to be.

Troubleshoot @pgsql answers

The answer is too generic

Make sure you're connected to the correct database and ask a schema-specific question instead of a broad PostgreSQL question.

The participant can't find the right objects

Confirm the object names in the Connections tree first, then include the schema or table names directly in your prompt.

I'm in Cursor and don't see @pgsql

The chat participant is registered only in Visual Studio Code. In Cursor, use Connect AI from the Connections tree to start Agent mode, or rely on the MCP server to expose the same tools to Cursor's native chat.

I want the AI to do more than answer

Use Agent mode when you want the AI to combine connection, schema inspection, queries, scripts, or data operations in one workflow.

Related content

  • Last updated on