Google Cloud Tools

Author: r3-yamauchi
Version: 0.0.1
Type: tool

English | Japanese

Overview

Google Cloud Tools is a lightweight collection of Dify tools that operate Google Cloud Storage (GCS) buckets and objects. With a single Service Account JSON key you can list buckets, browse objects, read or write data, delete assets, and mint signed URLs directly from a workflow.

The source code of this plugin is available in the GitHub repository.

Included Tools

Tool ID	Purpose
gcs_list_buckets	List buckets within the active project and return their region/storage class.
gcs_list_objects	Fetch up to 1000 objects from a bucket with optional prefix filtering.
gcs_create_bucket	Create a bucket with custom location, storage class, and uniform access.
gcs_read_object	Return the content of gs://bucket/object as text or Base64.
gcs_download_object	Download an object as a blob so downstream nodes can reuse it.
gcs_upload_object	Upload workflow files or text to GCS and optionally generate a signed URL.
gcs_delete_object	Remove an object and report the outcome as JSON.
gcs_generate_signed_url	Issue V4 signed URLs for GET/PUT/HEAD with a configurable TTL.

Requirements

1. Google Cloud project with the Cloud Storage JSON API enabled.
2. Service Account with at least roles/storage.objectAdmin.
3. The Service Account JSON key (PEM).
4. (Recommended) A bucket registered under the provider setting “Bucket Name (optional)” to avoid typing it for every tool.

Setup

1. Create the Service Account: Google Cloud Console → IAM & Admin → Service Accounts → New Service Account → grant Storage Object Admin (or a narrower custom role).
2. Download the JSON key and store it securely.
3. Register the plugin in Dify: Plugins → Add Plugin → upload the my_google_cloud_tools directory and fill in the credential form:
   • Service Account JSON – entire JSON key (optional; each tool form also exposes this field if you prefer per-call secrets).
   • Project ID (optional) – fallback when the JSON key omits project_id or when you need to target a different project.
   • Bucket Name (optional) – default bucket used whenever a tool input omits bucket_name.
   • Location (optional) – default bucket location (e.g., asia-northeast1) used by creation tools.
4. Validate by invoking gcs_list_objects or gcs_list_buckets in a test workflow to ensure authentication succeeds.

Usage Tips

• Leave a tool’s bucket_name blank to reuse the provider-level bucket.
• gcs_upload_object prioritizes input_file when both a file and text_content are supplied.
• Signed URL expirations (expires_in, presign_expiry) must stay between 60 and 604800 seconds.
• Use output_format=base64 in gcs_read_object when downstream nodes need binary-safe payloads.

Development & Testing

uv pip install -r my_google_cloud_tools/requirements.txt
pytest tests/test_provider_my_google_cloud_tools.py
pytest tests/tools/test_gcs_list_buckets.py
pytest tests/tools/test_gcs_list_objects.py
pytest tests/tools/test_gcs_download_object.py
pytest tests/tools/test_gcs_create_bucket.py

License

Apache License 2.0 — see LICENSE and NOTICE for details.