Google Cloud Tasks API
Browse all Google APIs✓ Official Vendor SpecCloud InfrastructureServerlessoauth216 EndpointsREST
For Agents
Create rate-limited task queues, enqueue HTTP or App Engine work items, and pause, resume, or purge queues on demand. Lets agents reliably hand off work to background workers without managing a queue cluster.
Use for: I need to create a Cloud Tasks queue with a max dispatch rate, Enqueue a task that POSTs to my worker endpoint, Pause a Cloud Tasks queue during an incident, Resume a paused Cloud Tasks queue
Not supported: Does not stream events, run cron schedules, or execute the task code itself — use for asynchronous, rate-limited dispatch of HTTP and App Engine targets only.
Cloud Tasks is a fully managed service for the asynchronous execution of distributed work items. The API exposes queue and task primitives so you can enqueue HTTP or App Engine targets, control delivery rate, set retry policy, and pause queues during incidents. It is the backbone for offloading long work from request handlers, smoothing traffic spikes, and decoupling producers from consumers without standing up a Kafka or Redis cluster.
Install Jentic One Beta
Connect the Cloud Tasks API to your agent
Jentic One is a self-hosted execution layer for AI agents. It lets your agent call the Cloud Tasks API, or any other public or private API you need. You set the rules, the agent never sees your credentials, and every call is logged.
Two steps, two machines. Install the instance in a safe environment, then register your agent from wherever it runs.
1
Step 1: Jentic One Host machine
# On the machine that will host your Jentic One instance:
curl -fsSL https://raw.githubusercontent.com/jentic/jentic-one/main/tools/install.sh | sh2
Step 2: Agent machine
# On the machine where your agent runs (keep this separate from the instance):
curl -fsSL https://raw.githubusercontent.com/jentic/jentic-one/main/tools/install.sh | sh
jentic register # connects your agent to your Jentic One instanceJentic One is in public beta. The setup above keeps your agent separate from the instance, which is what you want before using real credentials: an agent running as the same OS user as Jentic One can read its stored keys directly. Just evaluating? A single local install is fine to start. See the secure deployment guide for the tiers.
Capabilities
What an agent can do with Cloud Tasks API.
Create queues with configurable max dispatch rate, concurrency, and retry policy
Enqueue tasks that POST to HTTP endpoints with optional OIDC token authentication
Pause and resume queues during incidents to prevent thundering herds
Force-run queued tasks immediately for testing or manual replay
Purge all queued tasks when a backlog needs to be discarded
List and inspect tasks waiting in a queue
Use Cases
Patterns agents use Cloud Tasks API for, with concrete tasks.
★ Offload Long Work From Request Handlers
Hand off slow operations like PDF generation, third-party API calls, and bulk emails to a Cloud Tasks queue so the user-facing request returns in milliseconds. The API enqueues an HTTP target with a payload, retry policy, and optional delay, and Cloud Tasks invokes the worker with reliable at-least-once delivery and exponential backoff.
Enqueue a task that POSTs {"order_id": 1234} to https://workers.example.com/generate-invoice with a 5 minute delay and 5 retry attempts
Rate-Limited Outbound Calls
Use a Cloud Tasks queue to throttle calls to a third-party API that imposes strict QPS limits. Configure a max dispatch rate and concurrency on the queue, enqueue every outbound call as a task, and Cloud Tasks paces the workers so the upstream service is never overwhelmed even during spikes.
Create a queue with maxDispatchesPerSecond=10 and enqueue 5,000 tasks that each POST to the partner's webhook endpoint
Incident-Time Queue Pausing
Pause a queue during an outage so tasks accumulate without firing against a broken downstream service, then resume once the service is healthy. Cloud Tasks retains queued items while paused and continues to apply the queue's retry policy on resume, which keeps work in flight without losing items or hammering a recovering system.
Pause queue projects/acme/locations/us-central1/queues/email-out, wait for the downstream alert to clear, then resume the queue
Agent-Controlled Background Work
An AI agent can enqueue follow-up work for itself instead of holding a long synchronous loop. Through Jentic the agent searches for 'enqueue cloud task', loads the tasks.create schema, and submits a task that calls back into the agent's webhook with the next step's input.
Use Jentic to enqueue a task on queue email-out that POSTs the next chunk of work to the agent's resume webhook in 60 seconds
Key Endpoints
16 endpoints — cloud tasks is a fully managed service for the asynchronous execution of distributed work items.
METHOD
PATH
DESCRIPTION
GET
/v2/{+name}/locations
List Cloud Tasks regions for the project
POST
/v2/{+name}:pause
Pause dispatching from a queue
POST
/v2/{+name}:resume
Resume a paused queue
POST
/v2/{+name}:purge
Purge all tasks from a queue
POST
/v2/{+name}:run
Force-run a queued task immediately
GET
/v2/{+name}/locations
List Cloud Tasks regions for the project
POST
/v2/{+name}:pause
Pause dispatching from a queue
POST
/v2/{+name}:resume
Resume a paused queue
POST
/v2/{+name}:purge
Purge all tasks from a queue
POST
/v2/{+name}:run
Force-run a queued task immediately
Why Jentic?
Three things that make agents converge on Jentic-routed access.
Credential isolation
Google OAuth 2.0 refresh tokens for Cloud Tasks are stored encrypted in the Jentic vault. Agents receive scoped short-lived access tokens at execution time, and OIDC tokens for invoking private worker services are minted on-demand by the underlying service account.
Intent-based discovery
Agents search Jentic with intents like 'enqueue a task' or 'pause cloud tasks queue' and Jentic returns the matching queues or tasks operation with its request schema.
Time to first call
Direct integration with Cloud Tasks: 1-3 days for OAuth setup, queue tuning, and OIDC target wiring. Through Jentic: under 1 hour — search, load schema, execute against an existing queue.
Related APIs
Alternatives and complements available in the Jentic catalogue.
FAQs
Specific to using Cloud Tasks API through Jentic.
What authentication does the Cloud Tasks API use?
It uses Google OAuth 2.0 with the https://www.googleapis.com/auth/cloud-platform scope. Through Jentic the OAuth credentials live encrypted in the Jentic vault and the agent only sees a short-lived access token at execution time.
Can Cloud Tasks call private Cloud Run services?
Yes. Configure the task's HTTP target with an OIDC token from a service account that has run.invoker on the target service. Cloud Tasks adds the bearer token to the dispatched request so the worker can be locked down to authenticated callers.
What are the rate limits for the Cloud Tasks API?
Each queue has a configurable maxDispatchesPerSecond and maxConcurrentDispatches that govern how fast tasks are dispatched. The control-plane API for creating queues and enqueuing tasks has its own per-project quota that defaults to several thousand operations per minute.
How do I enqueue a background task through Jentic?
Search Jentic for 'enqueue cloud task', load the tasks.create schema, and execute the call against /v2/{parent}/tasks with the HTTP target URL, body, and headers. Jentic returns the schema so the agent does not need to parse Cloud Tasks' resource model by hand.
Is the Cloud Tasks API free?
Cloud Tasks includes a free tier of one million operations per month per billing account and charges per million operations after that. The API itself has no per-call markup.
How do I drain a queue without losing tasks?
Pause the queue with /v2/{name}:pause so new dispatches stop, let in-flight work complete, then resume or purge as needed. Purge removes all queued tasks irreversibly, so use it only when the queued items can be discarded.
GET STARTED