Amazon Redshift — Claude Code MCP server: what it does, how to install
Amazon Redshift — an MCP server for Redshift that discovers clusters and serverless workgroups, browses schemas and tables, and runs SQL. Read-only unless you opt out.
Apache-2.0 ★ 9,939 updated 2026-10-23
Open on GitHubWhat Amazon Redshift does
The server gives the agent tools for Redshift work: it finds provisioned clusters and serverless workgroups, walks databases, schemas, tables and columns, and runs SQL. Queries can go in parallel against the same database, and several calls can share one named transaction that you commit or roll back. A separate review checks a cluster or workgroup for health and configuration problems and returns a recommendation per finding.
Read-only is the default: one statement per call, writes rejected. ACCESS_MODE=read-write opens changes and deletes, and every statement that could alter something then waits for your approval. That prompt is the last point where you can stop it. An unsafe mode that skips the prompts stays off unless you turn it on.
Requirements: uv, Python 3.10 or newer, AWS credentials and a region. What actually guards the data is the database user's privileges, so create a dedicated user limited to the objects in play — an admin or superuser identity has no business behind a write-capable server. On a production cluster, confirm you have a snapshot you can restore first.
What it helps with
- list every cluster and serverless workgroup in the account
- inspect schemas and columns before rewriting a query
- run an analytical query from the editor without opening the console
- group statements into a named transaction and roll it back
- audit a cluster for health and configuration issues
Example prompts for the agent
- List the Redshift clusters and serverless workgroups I can reach, with their regions.
- Show the tables in the
analyticsschema and the columns oforders. - Run a read-only query that totals revenue per day for last month.
How to install Amazon Redshift
Install
The MCP registry has no launch command for it: see the repository README
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Amazon Redshift
How do I install the Amazon Redshift MCP server in Claude Code?
Add an entry to your MCP client config: command set to uvx, args set to awslabs.redshift-mcp-server@latest, and AWS_PROFILE plus AWS_DEFAULT_REGION under env. Install uv and Python 3.10 or newer first, then restart the client.
What does the Amazon Redshift MCP server do?
It discovers provisioned clusters and serverless workgroups, browses databases, schemas, tables and columns, executes SQL, and reviews a cluster for health and configuration issues.
Can the server change or delete data in Redshift?
Only when ACCESS_MODE=read-write is set. In that mode each statement that could alter something asks for approval, and without a transaction there is no undo. The database user's privileges remain the real safeguard, so skip admin accounts.