> ## Documentation Index
> Fetch the complete documentation index at: https://cool.computer/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Run a published service

> Saves and starts one managed service, then exposes its HTTP port after successful startup. On a published site with no VM yet it gives the computer one first, as exec does. On a computer whose URL serves published files, the files keep being served and the service answers what no file matches. The command must listen on the declared port; one still starting after 60 seconds is kept and answers 202. A later explicit exposure selection is independent of this process. Send an Idempotency-Key when a network failure might cause a retry.



## OpenAPI

````yaml https://www.cool.computer/openapi.json post /api/computers/{id}/service/run
openapi: 3.1.0
info:
  title: Cool Computers API
  version: 1.0.0
  description: >-
    Create and operate reachable Linux computers. Each computer keeps a stable
    name.cool.computer address, persistent files, and agent-ready command
    execution. REST requests use UTC-aligned one-minute limits: 600 requests per
    account identified by a recognized valid account bearer credential, a
    separate 600-request budget per computer-scoped credential, or 60 requests
    per source IP otherwise. Read-only GETs answered from API state spend a
    separate read budget reported as account-read or computer-read, so status
    polling cannot exhaust the budget that protects work and mutations. To bound
    authentication-provider traffic, at most 60 first-seen bearer values are
    checked per source IP in a minute; recognized valid credentials use their
    account or computer policy. Successful bearer authentication is reused for
    at most five seconds; transient provider failures are not cached. Non-exempt
    API responses include the structured RateLimit and RateLimit-Policy fields;
    request admission 429 responses use code api_request_rate_exceeded and also
    include Retry-After. Health, OpenAPI, webhooks, callbacks, SSE, and
    WebSocket endpoints are exempt from the REST request budget. Authenticated
    SSE connection attempts remain subject to the authentication safeguards and
    can receive the same typed 429 response.
servers:
  - url: https://api.cool.computer
    description: Canonical API origin
security: []
tags:
  - name: Authentication
    description: Obtain and manage bearer credentials.
  - name: Billing
    description: Inspect and manage the account's no-card trial or Pro plan.
  - name: Computers
    description: Create, inspect, and delete resident computers.
  - name: Execution
    description: Run commands on a computer.
  - name: Files
    description: Read and change a computer's persistent files.
  - name: Publishing
    description: >-
      Publish files to a computer's URL as immutable releases, with no VM. To
      show something, publish the files; to run something, use the computer
      (exec, terminal, services), which gets Linux on first use.
  - name: Services
    description: >-
      Manage processes and select the HTTP port exposed at a computer's stable
      address.
  - name: Email
    description: Read and send mail through a computer's inbox.
  - name: Network
    description: Control a computer's public visibility and outbound network policy.
  - name: Sharing
    description: Invite people to use a computer without granting owner controls.
  - name: Public
    description: Unauthenticated discovery endpoints.
externalDocs:
  description: Cool Computers agent registration guide
  url: https://www.cool.computer/auth.md
paths:
  /api/computers/{id}/service/run:
    post:
      tags:
        - Services
      summary: Run a published service
      description: >-
        Saves and starts one managed service, then exposes its HTTP port after
        successful startup. On a published site with no VM yet it gives the
        computer one first, as exec does. On a computer whose URL serves
        published files, the files keep being served and the service answers
        what no file matches. The command must listen on the declared port; one
        still starting after 60 seconds is kept and answers 202. A later
        explicit exposure selection is independent of this process. Send an
        Idempotency-Key when a network failure might cause a retry.
      operationId: runComputerService
      parameters:
        - $ref: '#/components/parameters/ComputerID'
        - $ref: '#/components/parameters/IdempotencyKey'
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/ServiceRunRequest'
      responses:
        '200':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: The configured service and current process state.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Service'
        '202':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: >-
            The service is still starting in the background: its process runs,
            but its port did not open within 60 seconds. It is kept, and its
            port is exposed once it listens; detail says what happens next.
          content:
            application/json:
              schema:
                allOf:
                  - $ref: '#/components/schemas/Service'
                  - type: object
                    required:
                      - detail
                    properties:
                      detail:
                        type: string
        '400':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: Bad Request.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
        '401':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: Unauthorized.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
        '402':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: >-
            Payment Required. The trial expired or Pro is inactive; inspect the
            problem code and resolution.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
        '403':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: Forbidden.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
        '404':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: Not Found.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
        '409':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: Conflict.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
        '413':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: Payload Too Large.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
        '429':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
            Retry-After:
              $ref: '#/components/headers/Retry-After'
          description: >-
            Too Many Requests. If the problem code is api_request_rate_exceeded,
            wait for Retry-After seconds; other problem codes describe a
            separate limit.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
        '500':
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: Internal Server Error.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
        default:
          headers:
            RateLimit:
              $ref: '#/components/headers/RateLimit'
            RateLimit-Policy:
              $ref: '#/components/headers/RateLimit-Policy'
          description: >-
            A machine-readable API error with a stable code and recovery
            instruction.
          content:
            application/problem+json:
              schema:
                $ref: '#/components/schemas/Problem'
      security:
        - bearerAuth: []
components:
  parameters:
    ComputerID:
      name: id
      in: path
      required: true
      description: Stable public computer identifier.
      schema:
        type: string
        minLength: 1
        example: cmp_123
    IdempotencyKey:
      name: Idempotency-Key
      in: header
      required: false
      description: >-
        A caller-generated key for safely retrying this side effect. The server
        retains the key and response for 24 hours. Reusing it with the same
        request replays the original response; reusing it with different input
        returns 409.
      schema:
        type: string
        minLength: 1
        maxLength: 128
        pattern: ^[!-~]+$
        example: req_01J6M6Y9TF9Q9QF6M1ZP8K4R2A
  schemas:
    ServiceRunRequest:
      type: object
      additionalProperties: false
      required:
        - command
        - port
      example:
        command: python3 -m http.server 3000
        port: 3000
      properties:
        command:
          type: string
          minLength: 1
        cwd:
          type: string
          description: >-
            Existing absolute directory used to start the service. Omit it to
            use /home/runtime/app, created if missing. /, /home and
            /home/runtime hold credentials and are refused with 400.
          default: /home/runtime/app
        port:
          type: integer
          description: >-
            HTTP port to expose after the command starts. Platform ports 3129,
            8080 and 18080 are reserved.
          minimum: 1
          maximum: 65535
          not:
            enum:
              - 3129
              - 8080
              - 18080
    Service:
      type: object
      additionalProperties: false
      required:
        - configured
        - status
      properties:
        configured:
          type: boolean
        status:
          type: string
          description: Current lifecycle or process status.
        public_url:
          type: string
          format: uri
        command:
          type: string
        cwd:
          type: string
        port:
          type: integer
          minimum: 1
          maximum: 65535
        exec_id:
          type: string
        pid:
          type: integer
          minimum: 1
        started_at:
          type: string
          format: date-time
        ended_at:
          type: string
          format: date-time
        exit_code:
          type: integer
        signal:
          type: string
        restart_policy:
          type: string
          const: none
        readiness:
          $ref: '#/components/schemas/ServiceReadiness'
        logs:
          $ref: '#/components/schemas/ServiceLogState'
    Problem:
      type: object
      additionalProperties: false
      required:
        - type
        - title
        - status
        - code
        - detail
        - resolution
      properties:
        type:
          type: string
          format: uri
        title:
          type: string
        status:
          type: integer
          minimum: 400
          maximum: 599
        code:
          type: string
          pattern: ^[a-z0-9_]+$
        detail:
          type: string
        resolution:
          type: string
        context:
          type: object
          additionalProperties: true
    ServiceReadiness:
      type: object
      additionalProperties: false
      required:
        - type
        - ready
        - continuous
      properties:
        type:
          type: string
          const: port_ready
        ready:
          type: boolean
        continuous:
          type: boolean
          const: false
    ServiceLogState:
      type: object
      additionalProperties: false
      required:
        - retention
        - bytes_written
        - bytes_dropped
        - dropped_frames
        - next_id
      properties:
        retention:
          type: string
          const: disk_ring
        bytes_written:
          type: integer
          format: int64
          minimum: 0
        bytes_dropped:
          type: integer
          format: int64
          minimum: 0
        dropped_frames:
          type: integer
          minimum: 0
        next_id:
          type: integer
          format: int64
          minimum: 0
  headers:
    RateLimit:
      description: >-
        The applied policy name, requests remaining after this request, and
        whole seconds until the current UTC-aligned window resets.
      schema:
        type: string
        pattern: >-
          ^"(account|account-read|computer|computer-read|source)";r=[0-9]+;t=([1-9]|[1-5][0-9]|60)$
        examples:
          - '"account";r=599;t=42'
          - '"computer-read";r=598;t=42'
    RateLimit-Policy:
      description: >-
        The selected policy name, quota, and window in seconds. Valid account
        credentials use account or account-read with q=600, computer credentials
        use computer or computer-read with q=600, and unauthenticated or
        unrecognized requests use source with q=60 for their source IP.
      schema:
        type: string
        pattern: >-
          ^("(account|account-read|computer|computer-read)";q=600|"source";q=60);w=60$
        examples:
          - '"account";q=600;w=60'
          - '"computer-read";q=600;w=60'
    Retry-After:
      description: >-
        Whole seconds to wait after an api_request_rate_exceeded response before
        retrying. Other 429 problem codes describe separate resource or provider
        limits and do not inherit this retry value.
      schema:
        type: integer
        minimum: 1
        maximum: 60
        examples:
          - 42
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      bearerFormat: WorkOS user or agent access token, or Cool Computers API key
      description: >-
        A Cool Computers API key, user access token, or claimed agent access
        token in the Authorization header. Follow
        https://www.cool.computer/auth.md for agent registration and
        https://www.cool.computer/authentication.md for direct user credentials.

````