deployment_service_create

Tool

deployment_service_create

Create a service inside an environment.

Create a service in an existing environment from a git branch, a container image, or a static build. Requires product-scoped write access. Choose persistent mode for a running service or scheduled mode for a job; mode cannot change after creation. Returns: `{ id, environmentId, name, slug, source, ... }` — a service object, not a deployment id. Typical sequence: deployment_service_create → deployment_deployment_list with the returned service id → deployment_deployment_get with a deployment id from that list. Fails when: scheduled mode is combined with containerPort, replicas, or scaleToZero. Replace the example environmentId with your environment's UUID.

Arguments

Name	Type	Required	Description
mode	string	—	persistent keeps a service running; scheduled runs a job. Cannot change after creation.default: persistent
name	string	•	Service name, 1–80 characters. The example uses api.
slug	string	—	URL slug, 1–40 characters.
canvas	object	—	Position on the environment canvas, with numeric x and y fields.
probes	object	—	Readiness and liveness checks. Each configured check requires a path starting with /; timing fields use seconds.default: {}
source	object	•	Source object: type is git, image, or static. Git requires branch; image requires registry and tag. The example uses the main branch; replace it with your branch and supply the repository details for your source.
command	string \| null	—	Shell command for a scheduled run, 1–4096 characters. Omit it or use null to run the image entrypoint.
tracing	string	—	Tracing instrumentation: off, node, or python. Applied when a deployment builds the workload.default: off
replicas	integer	—	Number of replicas, 0–50. Persistent mode only.
schedule	string \| null	—	Five-field UTC cron expression: minute hour day-of-month month day-of-week. Use null for a job that runs only on demand.
resources	object	—	CPU and memory limits, supplied as strings in limits.cpu and limits.memory.
containerPort	integer	—	Container port, 1–65535. Persistent mode only.
environmentId	string	•	UUID of the existing environment. Replace the example UUID with yours.
jobTimeoutSec	integer	—	Maximum duration of a scheduled run, in seconds, from 60 to 86400. A run exceeding this limit is terminated.
jobConcurrency	string	—	When a previous run is still active: forbid skips the next run, allow runs both, and replace terminates the previous run.

Try it

Create a service called api in my staging environment from the main branch of my repo.

When to use

Example

Ask your agent something like this:

Create a service called api in my staging environment from the main branch of my repo.

The agent will invoke deployment_service_create with these arguments:

{
  "environmentId": "11111111-1111-4111-8111-111111111111",
  "name": "api",
  "source": {
    "type": "git",
    "branch": "main"
  }
}