Amazon Redshift
Discover warehouse metadata and run bounded statements through the Redshift Data API
Connect Amazon Redshift when a project needs warehouse metadata or SQL results through the Redshift Data API. Swarmz fixes each connection to one serverless workgroup or provisioned cluster and signs every request on the server.
Availability
A Swarmz administrator must show Amazon Redshift under Platform config > Integrations and enable the data and analytics connector flag. Redshift does not require a Swarmz OAuth application.
Before you connect
Enable the Redshift Data API and create an IAM principal with the minimum actions the project needs. Do not use an AWS root access key. AWS documents Data API authorization.
Choose one deployment target:
- For Redshift Serverless, enter a workgroup name. Do not enter a cluster identifier or database user.
- For a provisioned cluster, enter a cluster identifier. Add a database user only when your authentication setup requires it.
Connect Amazon Redshift
- In the Swarmz project, open Settings > Integrations and select Amazon Redshift.
- Enter the AWS access key ID, secret access key, and optional session token.
- Enter the AWS region and Database.
- Enter
serverlessorprovisionedunder Deployment type. - Enter the matching workgroup or cluster field. Leave fields for the other deployment type empty.
- Select Save connection.
Swarmz signs a Data API request before saving. The target stays in server-held connection metadata; an AI operation cannot switch the request to another cluster or workgroup.
Available operations
| Operation | What it does | Limit per minute |
|---|---|---|
list_databases | Lists databases with an optional pattern | 30 |
list_schemas | Lists schemas with an optional pattern | 30 |
list_tables | Lists tables with optional schema and table patterns | 30 |
describe_table | Reads columns for one table | 30 |
execute_statement | Starts an asynchronous SQL statement | 10 |
describe_statement | Reads statement status | 30 |
get_statement_result | Reads one bounded result page | 30 |
cancel_statement | Cancels an active statement | 10 |
Executing or canceling a statement requires allow-once approval. SQL supports named parameters so the generated app does not need to concatenate user values into a query.
Ask the AI to use Amazon Redshift
- “List the schemas and build an orders dashboard from the warehouse tables.”
- “Run this parameterized customer query, poll it, and render the completed result.”
- “Cancel the long-running statement after showing me its status and ID.”
Verify Amazon Redshift
- List databases, schemas, and tables.
- Describe one known table and confirm its columns.
- Execute
select 1, poll the returned statement ID, then read the result. - Confirm the statement in the Redshift query history for the configured target.
An authorization error means the IAM policy, database grants, or target identity does not permit the operation. Rotate credentials in Swarmz before disabling the old IAM key. Select Remove and revoke the key to remove access.