Project structure reference

Overview

A declarative pipeline project uses the same folder layout whether it is connected to a GitHub repository, exported as a ZIP file, or both. This page describes the project-level files and the task-level files and folders that can appear in a declarative pipeline project.

Folder structure

my-pipeline-project/
  qtcp_project.yaml
  qtcp_bindings_definition.json
  bindings.json                 # export only, included only when binding values are requested
  qtcp_tasks/
    newTaskDefaults.yaml        # optional
    {taskName}/
      task.yaml
      transformationRules.yaml  # optional
      sourceSelection.yaml      # optional
      schedule.yaml             # optional
      model.yaml                # optional, transform task types only
      transformationDataFlows/  # optional, transform task types only
        {transformationDataFlowName}.yaml
      datasets/                 # optional, task-dependent
        {datasetName}.yaml

Project-level files

qtcp_project.yaml defines project-wide settings, such as the project name, platform type (for example, Snowflake or Databricks), and space assignment. This is the root file for the project and is always present.

qtcp_bindings_definition.json is the variable registry for the project. Every binding variable used in the YAML files must be declared here. The file lists variable names and is part of the project definition. Qlik Cloud uses this file to resolve binding references and expose environment-specific values through the Bindings API.

bindings.json contains the resolved, environment-specific values for variables declared in qtcp_bindings_definition.json. This file can be included in exported project packages when binding values are requested. It is always a JSON file, even when the project definition uses YAML.

Version-control commits do not include bindings.json. In the version-control workflow, binding values are stored in Qlik Cloud and resolved when you run Apply remote changes or use the Bindings API.

Task files

Each task lives in its own folder under qtcp_tasks/. The folder name matches the task name.

newTaskDefaults.yaml is optional and lives at the qtcp_tasks/ level. It sets project-wide default settings for each task type when new tasks of that type are created. The file is absent if no task-type defaults have been overridden.

task.yaml contains the core task configuration: task name, type, and settings that override defaults.

transformationRules.yaml is present when task-level transformation rules are defined. It holds rules that apply across all datasets in the task, such as global table or column renames.

sourceSelection.yaml defines which source tables or objects the task ingests. For tasks that read from a structured source, this file lists the selected tables.

schedule.yaml is present only when the task has a schedule defined. It contains the schedule type and cadence settings. If the task runs on demand only, this file is omitted.

model.yaml is present for transform task types. It contains the relationship definitions between datasets in the task. The file has a relationships array with one entry per defined relationship.

Dataset files

The datasets/ folder contains dataset files when dataset-level configuration is included for the task. On landing tasks, datasets without explicit rules can be omitted and regenerated from source selection metadata on import.

A dataset file contains the dataset properties and either a sourceTable section (for landing, lake landing, registered, and streaming landing task types) or an inputDatasets section (for all other task types), followed by any explicit rules.

Transformation data flow files

The transformationDataFlows/ folder is present for transform task types when the task contains transformation flows. Each flow is a separate YAML file named after the flow.