Skip to content

Getting started - using services

Run analyses instantly on ready-made services - no cloud setup, DevOps, or coding required. Run once or thousands of times, and add them to automation pipelines.

This guide walks you through using an example Twined service locally. The process for using a real one (deployed locally or in the cloud) is almost identical.

By the end, you will be able to use the Twined CLI to run an analysis on a data service, sending it input data and receiving output data.

Prerequisites

Before you begin, ensure you:

  • Are familiar with Python and/or the command line
  • Have the following tools installed:

Authentication

No authentication is needed to run the example data service. To authenticate for real data services, see authentication instructions.

Run your first analysis

Info

In Twined, sending input data to a service is called "asking a question". The service will run an analysis on the question and send back any output data - this is called called "receiving an answer".

Ask a question

The following command asks a question to the local example data service, which calculates the first n values of the Fibonacci sequence.

octue twined question ask example/service:latest --input-values='{"n": 10}'

Tip

To ask a question to a real data service, just specify its ID:

octue twined question ask some-org/a-service:1.2.0 --input-values='{"n": 10}'

from octue.twined.resources import Child

child = Child(
    id="example/service:latest",
    backend={
        "name": "GCPPubSubBackend",
        "project_id": "example",
    },
)

answer, question_uuid = child.ask(input_values={"n": 10})

Info

A child is a Twined service you ask a question to, in the sense of child and parent nodes in a tree. This only becomes important when services use other Twined services as part of their analysis, forming a tree of services.

Tip

To ask a question to a real data service, specify its ID and project ID e.g. some-org/real-service:1.2.0 instead of example/service:latest.

Receive an answer

The output is automatically written to the command line. It contains log messages followed by the answer as JSON:

[2025-10-28 15:36:52,377 | INFO | octue.twined.resources.example] Starting Fibonacci sequence calculation.
[2025-10-28 15:36:52,377 | INFO | octue.twined.resources.example] Finished Fibonacci sequence calculation.
{"kind": "result", "output_values": {"fibonacci": [0, 1, 1, 2, 3, 5, 8, 13, 21, 34]}, "output_manifest": null}

Tip

You can pipe the output JSON into other CLI tools or redirect it to a file:

# Format the result using the `jq` command line tool
octue twined question ask example/service:latest --input-values='{"n": 10}' | jq

# Store the result in a file
octue twined question ask example/service:latest --input-values='{"n": 10}' > result.json

answer

>>> {
    "kind": "result",
    "output_values": {"fibonacci": [0, 1, 1, 2, 3, 5, 8, 13, 21, 34]},
    "output_manifest": None,
}

Next steps

Success

Congratulations on running your first analysis! For additional information, check out the following resources: