PlugWiki
redshift-mcp-server logo

Amazon Redshift MCP Server: Setup for Claude, Cursor and Other Apps

Official · by Amazon Web ServicesLocal
GetPyPIawslabs.redshift-mcp-serverawslabs/mcp/tree/HEAD/src/redshift-mcp-server

Lets your AI assistant explore Amazon Redshift clusters and run read-only SQL against your data warehouse.

This AWS Labs server runs on your computer. It uses your AWS credentials to find your Redshift clusters and workgroups, browse their tables and run SQL.

It is read-only unless you set ACCESS_MODE to read-write. The README warns that read-write mode lets the model change and delete data, so leave it off unless you need it.

Good for

  • Analysts who want to ask questions of Redshift data in plain words
  • Data engineers browsing schemas and checking cluster health
  • Teams that need a query tool that is read-only by default

Works in

12 should work

Tap an app to see how to set it up and any known problems.

Install redshift-mcp-server

Claude Code

  1. 1

    Copy this prompt.

    Add the redshift-mcp-server server to Claude Code. Run this command:
    
    claude mcp add redshift -- uvx awslabs.redshift-mcp-server
    
    Then check that it loaded and tell me if I need to sign in or restart.
  2. 2

    Paste it into Claude Code. It shows the change and asks before running it.

Requires uv (Python).

Built from the package data we hold. Not yet tested by us.

Tools

  • Lists your Redshift clusters and serverless workgroups

    Finds every provisioned cluster and serverless workgroup you can see, with type, status, configuration and tags. The other tools need the identifier and type it reports.

    Try: “List all my Redshift clusters and workgroups.”

    Network · list_clusters

  • Lists the databases in a cluster

    Returns each database with its owner, type (local or shared) and access details.

    Try: “Which databases are on the analytics cluster?”

    Network · list_databases

  • Lists the schemas in a database

    Returns schema names, owners and types, including external schemas.

    Try: “Show the schemas in the dev database on analytics-cluster.”

    Network · list_schemas

  • Lists the tables and views in a schema

    Returns tables, views and external tables in a schema with their type and permissions.

    Try: “What tables are in the public schema of dev?”

    Network · list_tables

  • Shows a table's columns and types

    Returns each column with its data type, nullability, default and length limits.

    Try: “Show the columns of the users table.”

    Network · list_columns

  • Runs a SQL statement on a cluster

    Runs one SQL statement and returns the rows. By default it is read-only and rejects writes. With read-write mode on, it can change data and asks you to approve each risky statement first.

    Try: “How many customers signed up last month? Use the production cluster.”

    Network · execute_query

  • Checks a cluster for health and setup problems

    Runs diagnostic queries on system views and returns findings with a recommendation for each. The database user needs the sys:monitor role.

    Try: “Review my analytics cluster and tell me what to fix.”

    Network · review_cluster

Read from the source code

Permissions

low risk · 3 things to watch

Low risk

By default the assistant can browse and query your warehouse but cannot change it. Leave ACCESS_MODE at read-only and give the database user only the tables you want it to see.

What it can do

Watch out for

  • Read-write mode can change and delete data

    It is off by default. If you turn it on, use a dedicated database user with only the privileges the task needs, never an admin. github.com

  • Query results go to your AI provider

    Rows the assistant reads are sent to the model. Grant the database user access only to tables you can share. github.com

  • Logs can contain your SQL

    At debug level the server logs full statements. Keep LOG_LEVEL at its default and treat any log file as sensitive. github.com

Security advisories

No published security advisories as of Nov 1, 2026.

Technical details
Turning on read-onlyRead-only is the default. Setting the ACCESS_MODE environment variable to read-write turns writes on; any other value falls back to read-only.
Talks toAmazon Redshift and the Redshift Data API in your AWS region
Sign-in and keysUses your AWS credentials from a named AWS profile (AWS_PROFILE) or environment variables, and connects as the database user of that AWS identity.
Scopes
  • redshift:DescribeClusters

  • redshift-serverless:ListWorkgroups

  • redshift-data:BatchExecuteStatement

  • redshift-data:DescribeStatement

  • redshift-data:GetStatementResult

  • redshift-serverless:GetCredentials

  • redshift:GetClusterCredentialsWithIAM

Questions

Can it change or delete data in Redshift?

Not by default. It starts read-only and rejects writes. Only if you set ACCESS_MODE to read-write can it run statements that change data, and it then asks you to approve each one.

How do I turn on read-write mode?

Set the ACCESS_MODE environment variable to read-write. The README warns against it with an admin identity and advises a dedicated database user with narrow privileges.

Does it work with Redshift Serverless?

Yes. It finds both provisioned clusters and serverless workgroups, and the other tools take the type with the identifier.

Which AWS permissions does it need?

Describe and list calls for clusters and workgroups, the Redshift Data API actions to run and read statements, and the actions that get database credentials. The README lists them. You also need SELECT on the tables you want to query.

Why does review_cluster fail with a permission error?

The database user needs the sys:monitor role, or to be a superuser. The README gives the GRANT statement to run as a superuser.

Is there a limit on result size?

Yes. A result of more than 1,000 rows is refused rather than cut short, unless you change MAX_RESULT_ROWS. Ask for a LIMIT or an aggregate.

Information

updated Nov 2, 2026

Publisher
Amazon Web Services (official)
Package
awslabs.redshift-mcp-server
Registry
PyPI
Transport
stdio
License
Apache-2.0
Last commit
Oct 30, 2026
Open issues
143
Listed in
Docker MCP Catalog

GitHubDockerDocker MCP Catalog

Other AWS MCP servers

Sources

Docs, issue trackers and discussions this page is built from. Issues, quotes and findings above link to their own source.

Show all 2
  1. Amazon Redshift MCP Server README · github.com
  2. Redshift server source · github.com

Facts from the official MCP registry, GitHub, npm, PyPI and the Docker catalog. Updated Nov 2, 2026.