Open Settings → Data sources and click Add data source. Choose Amazon S3 (or an S3-compatible storage) and fill in the following fields:
| Field | Required | Description |
|---|
| Endpoint | ✅ | e.g. s3.amazonaws.com, or the endpoint of your S3-compatible storage (MinIO, DigitalOcean Spaces, …) |
| Access Key ID | ✅ | Access key ID of the IAM user |
| Secret Access Key | ✅ | Secret access key of the same user |
| Bucket Name | ✅ | The bucket meinGPT should read |
| Base Path | optional | Folder prefix, if only part of the bucket should be indexed |
Click Connect to create the source and start the first sync.
Amazon S3 runs as a Cloud data source; it needs no Outpost of your own, and this source type is not available for on-premise Outpost deployments.
meinGPT only needs read access to the bucket:
- Create an IAM user — AWS IAM Console↗
- Attach a policy — read access (
s3:GetObject, s3:ListBucket) to the target bucket
- Generate access keys — copy the access key ID and secret access key into the form above
For S3-compatible providers (MinIO, DigitalOcean Spaces, …), the same flow applies at that provider instead of AWS IAM.
- meinGPT connects to the configured bucket and syncs objects below the base path.
- Later runs only process changes.
- For large buckets, start with a narrow base path for a safer rollout.