Built by Metorial, the integration platform for agentic AI.
List Compute Engine GPU and accelerator types available in a zone, including deprecation and card-limit details.
List instances in a Compute Engine managed instance group with instance status, intended template version, and current group action.
Cancel a running BigQuery job. The cancellation is best-effort; the job may still complete before the cancellation takes effect.
List global Compute Engine snapshots in the configured project with source and storage details.
Update a BigQuery table's metadata including friendly name, description, schema (add new columns), expiration, and labels.
Update an existing BigQuery dataset's metadata, including its friendly name, description, labels, and default expiration settings.
Retrieve detailed metadata for a BigQuery table, including its schema, row count, size, partitioning configuration, and clustering settings.
Stop a running Compute Engine VM and return the zone operation for asynchronous completion.
List BigQuery jobs in the project. Jobs include queries, loads, exports, and copy operations. Filter by state, time range, or parent job.
Retrieve detailed information about a specific BigQuery job including its configuration, status, and execution statistics.
Delete a Compute Engine VM and return the zone operation for asynchronous completion.
Get a Compute Engine reservation name, ID, creation timestamp, zone, status, commitment links, and targeting requirement.
List Compute Engine machine types available in a zone, including vCPU, memory, disk, accelerator, and architecture details.
List all datasets in the configured BigQuery project. Returns dataset IDs, friendly names, locations, and labels. Use the **filter** parameter to narrow results.
Get a Compute Engine VM name, ID, status, machine type, creation timestamp, and guest accelerators.
List the Compute Engine reservations attached to a regional commitment, including machine shape, accelerators, sharing, commitment links, and resource status.
List Compute Engine disks in a zone with basic identity, size, type, status, and attachment timestamps.
Delete a BigQuery dataset. By default, the dataset must be empty. Set **deleteContents** to true to also delete all tables and views within it.
List Compute Engine VMs in a zone with basic status, machine type, creation time, and accelerator details.
Read rows directly from a BigQuery table without running a query job. Useful for quickly inspecting table contents. For complex filtering or aggregation, use **Execute SQL Query** instead.
Permanently delete a BigQuery table or view. This action is irreversible.
Load data from Google Cloud Storage into a BigQuery table. Supports CSV, JSON (newline-delimited), Avro, Parquet, ORC, Datastore, and Firestore export formats. Creates an asynchronous load job and returns the job status.
Run a read-only GoogleSQL query against BigQuery. The exact SQL is dry-run first and executes only when BigQuery classifies it as a SELECT statement. Scripts, procedures, DML, and DDL are rejected.
Hard-reset a Compute Engine VM and return the zone operation for asynchronous completion.
Get a Compute Engine managed instance group name, ID, instance template, base name, target sizes, status, and creation timestamp.
List the complete Compute Engine attached-disk configurations for a VM instance.
List Compute Engine committed-use commitments in a region with status, plan, machine series, resources, and term timestamps.
Create a new BigQuery table, view, or materialized view. Supports defining schemas with nested/repeated fields, time or range partitioning, and clustering. To create a view, provide the **viewQuery** parameter; for a materialized view, provide **materializedViewQuery**.
Create a Compute Engine VM. Defaults to e2-medium and the debian-cloud debian-12 image family. Returns a zone operation for asynchronous completion.
Run a GoogleSQL (standard SQL) query against BigQuery. Supports SELECT, DML (INSERT, UPDATE, DELETE, MERGE), and DDL (CREATE, ALTER, DROP) statements. The query is submitted as a job, polled for completion, and results are returned. Parameterized queries are supported for safe value interpolation. You can optionally write results to a destination table.
Get a Compute Engine zone operation status, target, progress, request/start/end timestamps, errors, warnings, and HTTP failure details for polling asynchronous VM changes.
List Compute Engine reservations in a zone with basic identity, status, commitment links, and targeting requirements.
List Compute Engine images. By default aggregates the configured project and the standard public image projects; set imageProject to page through a single project such as debian-cloud or windows-cloud.
Get a Compute Engine disk name, ID, description, size, type, status, and attach or detach timestamps.
Get full details for a Compute Engine reservation, including machine shape, accelerators, local SSDs, aggregate resources, commitments, sharing, and resource status.
List Compute Engine managed instance groups in a zone with their templates, target sizes, status, and creation timestamps.
Export a BigQuery table to Google Cloud Storage as CSV, JSON, or Avro. Creates an asynchronous extract job. Use wildcards in the destination URI for sharded exports of large tables (e.g., gs://bucket/file-*.csv).
List global Compute Engine instance templates with their names, IDs, descriptions, machine types, regions, and creation timestamps.
Change the machine type of a stopped Compute Engine VM. The instance must be stopped before invoking this tool.
Get the type, size, provisioned IOPS, provisioned throughput, physical block size, storage pool, and access mode of a Compute Engine disk.
Retrieve detailed information about a specific BigQuery dataset, including its schema, access controls, creation time, and configuration.
Get the full instance properties of a global Compute Engine instance template, including disks, networks, metadata, scheduling, service accounts, accelerators, and security settings.
List all tables, views, and materialized views in a BigQuery dataset. Returns table IDs, types, creation times, and expiration info.
Get a regional Compute Engine commitment name, ID, status, plan, type, resources, and creation, start, and end timestamps.
Start a stopped Compute Engine VM and return the zone operation for asynchronous completion.
Create a new BigQuery dataset. A dataset is a top-level container for tables, views, and routines. Once created, its location cannot be changed.
Get a global Compute Engine instance template name, ID, description, machine type, region, and creation timestamp.
Create a new BigQuery routine: user-defined function (UDF), stored procedure, or table-valued function. Routines can be written in SQL or JavaScript.
Retrieve full details of a specific Cloud Function including its build configuration, service configuration, event trigger settings, state, and deployment URL. Use a short function name or a fully qualified resource name.
Send a push notification or data message via Firebase Cloud Messaging (FCM). Target individual devices by token, or broadcast to topics or conditions. Supports platform-specific configuration for Android, iOS (APNs), and Web.
List available runtime environments for Cloud Functions, including their lifecycle status (supported, deprecated, decommissioned). Useful for choosing a runtime when creating or updating functions.
Create, update, delete, or retrieve a Firebase Authentication user. Supports managing user properties including email, password, display name, phone number, photo, email verification status, and account disabled state.
List Firebase Android, iOS, and Web apps in a project, retrieve a specific app, or download its SDK configuration artifact. Use this to discover app IDs, package or bundle identifiers, web config, and native google-services files.
Permanently delete an object from a Cloud Storage bucket. Optionally specify a generation number to delete a specific version of a versioned object.
Delete a recognizer configuration. The recognizer enters a deleted state and is eventually purged.
Retrieve detailed information about a specific BigQuery routine including its definition, arguments, return type, and language.
Get or set IAM policies on a Cloud Storage bucket. Use **action: "get"** to retrieve the current policy, or **action: "set"** to replace the entire policy with the provided bindings. Use **action: "add_binding"** or **action: "remove_binding"** to modify individual bindings without replacing the full policy.
Generate a signed URL for uploading function source code to Cloud Storage. The returned URL and storage source should be used when creating or updating a function. Upload your source code archive (ZIP) to the signed URL before referencing it in a create or update call.
Analyzes an image to determine its dominant colors, returning RGB values, coverage fraction, and relevance scores. Useful for color palette extraction, design workflows, and image categorization by color.
List, create, or delete Pub/Sub notification configurations on a Cloud Storage bucket. Notifications deliver event messages to a Pub/Sub topic when objects are created, updated, deleted, or archived.
Detects and localizes multiple objects in an image, returning each object's name, confidence score, and bounding box coordinates. Useful for understanding object positions and spatial relationships within an image. Object names are returned in English only.
Create, update, or delete a Cloud Storage bucket. When creating, specify a name and optionally a location, storage class, and versioning. When updating, provide any fields to modify. When deleting, the bucket must be empty.
Stream rows into a BigQuery table using the streaming insert API. Rows are available for querying almost immediately. Each row is a JSON object matching the table schema. Optionally provide an insertId per row for best-effort deduplication.
Convert text or SSML into natural-sounding speech audio using Google Cloud Text-to-Speech. Returns base64-encoded audio data in the requested format. Supports multiple voice types including Standard, WaveNet, Neural2, Studio, and Chirp 3 HD voices. Customize pitch, speaking rate, and volume.
Create, read, update, or delete a document in Cloud Firestore. Supports setting document fields with automatic type encoding and specifying update masks for partial updates.
Get an object's metadata and optionally download its content from a Cloud Storage bucket. By default only returns metadata; set **includeContent** to true to return a binary-safe downloadable file.
Publish an updated Firebase Remote Config template or roll back to a previous version. When publishing, provide the full template with parameters, conditions, and the current ETag. Use the "Get Remote Config" tool first to retrieve the current ETag.
Get details of a specific recognizer configuration, including its model, languages, default recognition config, and status.
Check the status of a long-running operation such as function creation, update, or deletion. Provides the current state and any error or success details.
Copy an object within or between Cloud Storage buckets. The source object remains unchanged. Can also be used to rename or move an object by setting **deleteSource** to true.
List objects in a Cloud Storage bucket. Supports filtering by name prefix, delimiter-based folder simulation, and listing all object versions.
Update an existing recognizer configuration. Modify the display name, model, or language codes of a previously created recognizer.
Look up a Firebase Authentication user by email address or phone number. Useful for finding a user when you don't know their user ID.
Extracts text from images using optical character recognition (OCR). Supports two modes: standard text detection for photos and general scenes, and document text detection optimized for dense text, documents, and handwriting. Returns the full extracted text along with individual text blocks and their positions.
Create and deploy a new Cloud Function. Configure the runtime, entry point, source code location, memory, timeout, environment variables, and event triggers. Source code must be uploaded to a Cloud Storage bucket or referenced from a Cloud Source Repository. Returns a long-running operation that can be polled for completion.
Read, write, push, update, or delete data at any path in the Firebase Realtime Database. Supports querying with ordering, filtering, and pagination. Requires the **databaseUrl** to be configured in the project settings.
Detects faces in an image and returns detailed attributes including emotional expression likelihoods (joy, sorrow, anger, surprise), facial orientation angles, detection confidence, and whether headwear or blur is present. Includes bounding box coordinates for each detected face.
Searches the web for information related to an image. Returns matching web entities, pages containing the image, visually similar images, and best-guess labels describing the image content. Useful for reverse image search, finding image sources, and understanding web presence.
Create a named recognizer configuration for Speech-to-Text v2. A recognizer stores default settings like model, language, and recognition features so they don't need to be repeated in every transcription request.
Update an existing Cloud Function's configuration. Modify runtime settings, memory, timeout, environment variables, labels, description, scaling limits, ingress/egress settings, and more. Only the fields you provide will be updated. Returns a long-running operation.
Retrieve the current Firebase Remote Config template including all parameters, conditions, and parameter groups. Also supports listing version history for audit and rollback purposes.
Generate a signed URL for downloading the source code of a deployed Cloud Function. The URL is time-limited and can be used to retrieve the function's source archive.
Query or list documents from a Cloud Firestore collection. Supports structured queries with field filters, ordering, and pagination. Use the simple list mode for browsing documents or the query mode for filtered searches.
List, upload, get metadata, delete, or copy objects in Firebase Cloud Storage. Supports prefix-based listing for browsing folder-like structures, fetching download URLs, and copying objects between paths.
List available Text-to-Speech voices. Optionally filter by language code to find voices for a specific language. Returns voice names, genders, supported languages, and native sample rates.
List all recognizer configurations in the configured project and region. Returns recognizer names, models, languages, and status.
List Cloud Storage buckets in the configured project. Supports filtering by name prefix and pagination for large result sets.
Update metadata on an existing object in a Cloud Storage bucket. Supports changing content type, content disposition, cache control, custom metadata, and object holds.
Recognizes logos of popular brands within an image. Returns the brand name, confidence score, and bounding box for each detected logo. Useful for brand monitoring, market analysis, and detecting brand presence in social media images.
Analyzes an image for explicit or inappropriate content across five categories: adult, spoof, medical, violence, and racy. Returns a likelihood rating for each category. Useful for content moderation and filtering.
List Cloud Functions in a Google Cloud project. Returns a summary of each function including its state, runtime, URL, and labels. Supports filtering and pagination, and can query across all regions or a specific region.
List user-defined functions (UDFs), stored procedures, and table-valued functions in a BigQuery dataset.
Subscribe or unsubscribe device tokens to/from FCM topics. Batch subscribe or unsubscribe multiple devices at once.
Validates, standardizes, and geocodes a postal address. Accepts either structured address fields or unstructured address lines and returns a validated, formatted address along with component-level validation details, geocode coordinates, and a Google Place ID. For US/Puerto Rico addresses, optionally enables **USPS CASS processing** to obtain ZIP+4 codes, delivery point codes, and carrier route information. When re-validating a previously validated address, provide the `previousResponseId` to improve accuracy.
Get or set lifecycle management rules on a Cloud Storage bucket. Lifecycle rules automate storage class transitions, object deletion, and cleanup of incomplete uploads based on configurable conditions.
Start an asynchronous batch transcription of one or more audio files stored in Google Cloud Storage. Returns a long-running operation that can be monitored using the Get Operation tool. Suitable for audio files longer than 1 minute (up to 8 hours). Results can be written to a GCS output location or returned inline when the operation completes.
Upload a new object to a Cloud Storage bucket or overwrite an existing one. Provide the object name (path) and its text content. Optionally attach custom metadata key-value pairs.
Get or set the IAM policy for a Cloud Function. Use this to control who can invoke the function or manage its configuration. To make a function publicly accessible, grant the **roles/run.invoker** role to **allUsers**.
Performs multiple Vision API detection features on a single image in one request. Select any combination of features (labels, objects, faces, landmarks, logos, text, safe search, image properties, crop hints, web detection) to analyze an image comprehensively. More efficient than making separate calls for each feature.
Update BigQuery ML model metadata such as friendly name, description, expiration time, and labels. To change model training or prediction behavior, run SQL with **execute_query** instead.