How do I set up a Google Cloud bucket?
Use a Google Cloud project, a private bucket, and a service account dedicated to Transloadit. You need permission to create the bucket, manage its IAM access, and create a service account and its JSON key. Google Cloud storage and Transloadit processing charges apply.
1. Create a private bucket
Follow Google's bucket creation guide. Use uniform bucket-level access so that IAM controls access rather than individual object ACLs. Do not grant public access. Keep the bucket name and the project ID for the next steps.
2. Give a service account access to this bucket
Create a dedicated service account and grant it the Storage Object Creator role on this bucket, not your entire project. That role can create objects, but cannot read, delete, or overwrite existing objects. The example below uses the Robot's default unique output paths. If you choose fixed paths or add import Steps, review the permissions needed for that workflow instead of granting broad administrator access.
This Transloadit connection uses a service account's JSON key file. Follow Google's key creation guide and download a JSON key for the dedicated service account. Keep it private. If your organization prohibits service account keys, consult your administrator rather than bypassing that policy. An OAuth client file or an HMAC access key is not a replacement for this JSON key.
3. Save Template Credentials
Follow How to create Template Credentials and select Google Cloud Storage as the service. Fill in:
| Console field | Value |
|---|---|
| Google Project ID | Your Google Cloud project ID. |
| Google Cloud Storage Bucket | The bucket name, without a gs:// prefix. |
| Google JSON Key File | The service account JSON key file you downloaded. |
Give the credential a recognizable name. Use that name in the Template below, not the JSON file contents or the service account email address.
4. Test an export
Create a Template with these instructions, replacing
YOUR_GOOGLE_CLOUD_CREDENTIALS with the credential name you saved:
{
"steps": {
":original": {
"robot": "/upload/handle"
},
"exported": {
"robot": "/google/store",
"use": ":original",
"credentials": "YOUR_GOOGLE_CLOUD_CREDENTIALS",
"acl": null
}
}
}
Keep "acl": null: it omits the object ACL and lets the bucket's IAM policy control access.
Omitting acl instead uses the Robot's public-read default, which is incompatible with uniform
bucket-level access.
Run the Template with a small test upload. After the Assembly completes, inspect the export result and confirm the object exists using your own authenticated Google Cloud Console access. A plain result URL does not grant anonymous access to a private object. Do not make the bucket public just to test that URL.
An Assembly can fail if an automatic upload attempt creates the object but loses the confirmation: an automatic retry cannot replace that object with the Storage Object Creator role. Inspect the reported destination and the Assembly error before changing permissions. An existing object alone does not confirm that the complete export succeeded. To retry manually, start a new Assembly with a new upload and the example's default unique output paths instead of adding overwrite or delete permissions.
More
- See Template documentation for Template setup and security.
- See the 🤖/google/store documentation for paths, permissions, and other export options.