google-cloud-platform-api

name	google-cloud-platform-api
description	Explore BigQuery and Cloud Storage through bounded inspection and cost-controlled read tools.

Google Cloud Platform API

Use the local google_cloud_platform_api MCP server for Google Cloud inspection and analysis.

Operating contract

• Start with gcp_context to confirm the active project and defaults.
• No mutation tool is available; never claim that a cloud write was performed.
• Discover resources before querying them; do not guess identifiers.
• Follow next_cursor only when more discovery results are needed.
• Never reuse a cursor with a different dataset, bucket, or prefix.
• Prefer metadata inspection before reading object content.
• Treat bigquery_query as billable even though it cannot mutate data.

BigQuery workflow

1. bigquery_list_datasets
2. bigquery_list_tables
3. bigquery_table_schema
4. draft a single selective query
5. bigquery_query with an explicit acceptable maximum_bytes_billed

The query tool dry-runs first and rejects estimates above the billing ceiling. Use fully qualified table names, select only needed columns, include restrictive filters, and add a LIMIT whenever practical. A successful result reports estimated, processed, and billed bytes plus job metadata.

Cloud Storage workflow

1. gcs_list_objects
2. gcs_object_metadata
3. gcs_read_text

Read only plausibly textual objects. Start with a small max_bytes value. The server uses a provider byte range and reports whether more content exists.

Errors

Use the returned machine-readable code and guidance. Retry only when retryable is true. Permission failures should be resolved with the capability's documented IAM permissions; quota and availability failures should use bounded exponential backoff.