---
metadata:
  - name: generator
    content: Diplodoc Platform v5.62.0
alternate:
  - https://sourcecraft.dev/portal/docs/en/sourcecraft/ci-cd-ref/data-between-tasks.md
  - https://sourcecraft.dev/portal/docs/ru/sourcecraft/ci-cd-ref/data-between-tasks.md
  - href: https://sourcecraft.dev/portal/docs/en/sourcecraft/ci-cd-ref/data-between-tasks.md
    type: text/markdown
    title: Markdown version
  - href: https://sourcecraft.dev/portal/docs/en/llms.txt
    rel: describedby
title: Providing data between CI/CD tasks in SourceCraft
description: This page explains how to provide data between CI/CD tasks in SourceCraft.
---
> **Documentation Index:** Fetch the complete configuration index at https://sourcecraft.dev/portal/docs/en/llms.txt


# Providing data between CI/CD tasks in SourceCraft

In SourceCraft [CI/CD](https://sourcecraft.dev/portal/docs/en/sourcecraft/concepts/ci-cd.md), you can provide data between [tasks](https://sourcecraft.dev/portal/docs/en/sourcecraft/ci-cd-ref/tasks.md). Follow these steps:
1. In the source task, in the [cube](https://sourcecraft.dev/portal/docs/en/sourcecraft/ci-cd-ref/cubes.md) script, write the output data in `key=value` format to the [predefined](https://sourcecraft.dev/portal/docs/en/sourcecraft/ci-cd-ref/predefined-variables.md) environment variable named `SOURCECRAFT_OUTPUT`. Here is an example:

    ```yaml
    workflows:
      tasks-outputs-workflow:
        tasks:
          - name: producer
            cubes:
              - name: build
                script:
                  - |
                    echo "BUILD_RESULT=hello-world" >> $SOURCECRAFT_OUTPUT
                    echo "BUILD_VERSION=1.0.0" >> $SOURCECRAFT_OUTPUT
            ...
    ```

1. In the same task, in the `outputs` section, declare the list of variables you want to provide in this format: `<variable>: ${{ cubes.<cube_name>.outputs.<key_from_cube> }}`. Here is an example:

    ```yaml
    workflows:
      tasks-outputs-workflow:
        tasks:
          - name: producer
            ...
            outputs:
              PRODUCER_RESULT: ${{ cubes.build.outputs.BUILD_RESULT }}
              PRODUCER_VERSION: ${{ cubes.build.outputs.BUILD_VERSION }}
    ```

1. In the target task, add a dependency on the source task to the `needs` list in this format: `needs: [<task_name>]`. Here is an example:

    ```yaml
    workflows:
      tasks-outputs-workflow:
        tasks:
          - name: producer
            ...
          - name: consumer
            needs: [producer]
            ...
    ```

    <!-- source: en/_includes/sourcecraft/tasks-needs-restrictions.md -->
    {% note warning %}

    You can use the `needs` parameter only within the `workflows:tasks` section. You cannot use it in the separate `tasks` section.

    In the `needs` field, you can only reference tasks of the workflow within which the current task is executed. Circular dependencies are not allowed.

    {% endnote %}
    <!-- endsource: en/_includes/sourcecraft/tasks-needs-restrictions.md -->

1. In the cube of the dependent task, retrieve the data using this expression: `${{ tasks.<task_name>.outputs.<key_from_task> }}`. You can use this expression in scripts, environment variables, and the `outputs` section. Here is an example:

    ```yaml
    workflows:
      tasks-outputs-workflow:
        tasks:
            ...
          - name: consumer
            ...
            cubes:
              - name: consume
                env:
                  VERSION: ${{ tasks.producer.outputs.PRODUCER_VERSION }}
                script:
                  - echo "${{ tasks.producer.outputs.PRODUCER_RESULT }}"
                  - echo "$VERSION"
    ```

    {% note warning %}

    In a dependent task, you can retrieve output values only from tasks explicitly listed in `needs`.

    {% endnote %}

1. To forward data further down the dependency chain, declare a transit variable in the `outputs` section of an intermediate task. Here is an example:

    ```yaml
    workflows:
      tasks-outputs-workflow:
        tasks:
          - name: producer
            outputs:
              PRODUCER_RESULT: ${{ cubes.build.outputs.BUILD_RESULT }}
            ...
          - name: consumer
            needs: [producer]
            outputs:
              FORWARDED_RESULT: ${{ tasks.producer.outputs.PRODUCER_RESULT }}
            ...
          - name: next-consumer
            needs: [consumer]
            ...
    ```

## Example of a workflow with provision of data between tasks {#example}

In the example below:
* Data in the `PT_RESULT` variable from `producer-task` is provided to and used in `consumer-producer-task` and `consumer-task-2`.
* Data in the `CPT_RESULT1` and `CPT_RESULT2` variables from `consumer-producer-task` is provided to and used in `consumer-task-1` and `consumer-task-2`.

```yaml
workflows:
  tasks-outputs-workflow:
    checkout:
      enabled: false
    tasks:
      - name: producer-task
        # Section defining which data can be used by other tasks
        outputs:
          # Variable to write the task result to
          # The variable value must point to the key provided in the cube to the SOURCECRAFT_OUTPUT variable
          # The variable name in task `outputs` may not coincide with the key provided in the cube
          PT_RESULT: ${{ cubes.produce.outputs.PRODUCER_TASK_RESULT }}
        cubes:
          - name: produce
            script:
              # Providing the cube's output data to the SOURCECRAFT_OUTPUT environment variable in `key=value` format
              # Values can contain any characters, including special ones
              - echo "PRODUCER_TASK_RESULT=Result of producer-task — '🚀'" >> $SOURCECRAFT_OUTPUT

      - name: consumer-producer-task
        # Specifying a dependency on `producer-task` for access to its results
        needs: [producer-task]
        # `outputs` can be defined in any task, including a dependent one
        outputs:
          CPT_RESULT1: ${{ cubes.consume-produce-1.outputs.CPT_RESULT1 }}
          CPT_RESULT2: ${{ cubes.consume-produce-2.outputs.CPT_RESULT2 }}
        cubes:
          - name: consume-produce-1
            env:
              UPSTREAM_RESULT: ${{ tasks.producer-task.outputs.PT_RESULT }}
            script:
              - echo "$UPSTREAM_RESULT"
              - echo "CPT_RESULT1=$UPSTREAM_RESULT and the first result of consumer-producer-task — '⭐'" >> $SOURCECRAFT_OUTPUT
          - name: consume-produce-2
            script:
              - LOCAL_VAR="${{ tasks.producer-task.outputs.PT_RESULT }}"
              - echo "${LOCAL_VAR}"
              - echo "CPT_RESULT2=Second result of consumer-producer-task — '🔥'" >> $SOURCECRAFT_OUTPUT
          - name: consume
            script:
              - echo "${{ tasks.producer-task.outputs.PT_RESULT }}"

      - name: consumer-task-1
        needs: [consumer-producer-task]
        cubes:
          - name: consume
            script:
              # Only `consumer-producer-task` is specified in `needs`; therefore, only its data is available
              # Data from `producer-task` is not available
              - echo "${{ tasks.consumer-producer-task.outputs.CPT_RESULT1 }}"
              - echo "${{ tasks.consumer-producer-task.outputs.CPT_RESULT2 }}"

      - name: consumer-task-2
        needs: [producer-task, consumer-producer-task]
        cubes:
          - name: consume
            script:
              # `producer-task` and `consumer-producer-task` are specified in `needs`;
              # therefore, data from both tasks is available
              - echo "${{ tasks.consumer-producer-task.outputs.CPT_RESULT1 }}"
              - echo "${{ tasks.consumer-producer-task.outputs.CPT_RESULT2 }}"
              - echo "${{ tasks.producer-task.outputs.PT_RESULT }}"
```

## Limitations {#limits}

* The size of a single `KEY=VALUE` entry must not exceed 64 KB (65,536 bytes).
* The total size of the names and values of all output data of a single task must not exceed 1 MB. If the limit is exceeded, the task terminates with an error, even if all cubes are completed successfully. To provide larger data, use [artifacts](https://sourcecraft.dev/portal/docs/en/sourcecraft/ci-cd-ref/cubes.md#cubes-with-artifacts).
* If the `${{ cubes.<cube_name>.outputs.<key_from_cube> }}` or `${{ tasks.<task_name>.outputs.<key_from_task> }}` expression references a non-existent value, the task terminates with an error.

{% note tip %}

To use task output values on a [self-hosted worker](https://sourcecraft.dev/portal/docs/en/sourcecraft/concepts/workers.md#self-hosted-workers), [update](https://sourcecraft.dev/portal/docs/en/sourcecraft/operations/self-hosted-worker.md#update-worker) the worker executable to version 0.14.2 or higher.

{% endnote %}

#### Useful links {#see-also}

* [Tasks](https://sourcecraft.dev/portal/docs/en/sourcecraft/ci-cd-ref/tasks.md)
* [Providing environment variables from one cube to another in the SourceCraft CI/CD](https://sourcecraft.dev/portal/docs/en/sourcecraft/ci-cd-ref/env-vars-between-cubes.md)
* [Predefined environment variables](https://sourcecraft.dev/portal/docs/en/sourcecraft/ci-cd-ref/predefined-variables.md)
* [Continuous integration and continuous deployment in SourceCraft](https://sourcecraft.dev/portal/docs/en/sourcecraft/concepts/ci-cd.md)