Amazon Redshift
Query Redshift read-only. Connect with a host and password, or with AWS credentials if there is no route to the cluster.
What you will be asked for
- Host
- Port
- Database
- Username
- Encryption
- PasswordSecret
- Cluster identifier
- Serverless workgroup
- AWS region
- Access key ID
- Secret access keySecret
Where to find these
Worclaude queries Redshift live and read-only — nothing is copied into the
app, and every query it writes is a single SELECT.
1. Host and port
In the AWS console: Redshift → Clusters → your cluster → General information → Endpoint. It looks like:
my-cluster.abc123xyz.eu-west-1.redshift.amazonaws.com:5439/dev
└──────────────── host ────────────────────────────┘ └port┘ └db┘
Copy the three parts into the three fields. Port is almost always 5439.
2. Database
The last segment of that endpoint — often dev, prod or your team's name.
3. Username and password
Ask your data team for a read-only user. Do not use your own admin credentials, and do not reuse the cluster's master user: it can drop tables, and a question typed into a chat box should never be able to.
If you are the one creating it:
CREATE USER worclaude_ro PASSWORD 'a-long-random-string';
GRANT USAGE ON SCHEMA analytics TO worclaude_ro;
GRANT SELECT ON ALL TABLES IN SCHEMA analytics TO worclaude_ro;
ALTER DEFAULT PRIVILEGES IN SCHEMA analytics
GRANT SELECT ON TABLES TO worclaude_ro;
The last line matters — without it the user cannot read tables created after today, and the source quietly goes stale.
4. Network access
Redshift refuses connections from outside its security group. Worclaude connects
from this deployment's egress address, so that address has to be allowed in:
Cluster → Properties → Network and security → VPC security group
→ Inbound rules → allow TCP 5439 from it. Your infrastructure team will
know the address; a connection that hangs rather than refusing is almost always
this.
5. SSL mode
Leave it on require unless you have been told otherwise.
About a minute, once you have the credentials.