Getting Started

This guide walks you through installing Kuristo, writing your first workflow, and running it locally.

Installation

Clone the repository and install it from source:

git clone https://github.com/andrsd/kuristo.git
cd kuristo
pip install .

Basic Workflow

Kuristo workflows are written in YAML. Here’s a minimal example:

jobs:
  single-case:
    - name: simple test
      steps:
        - run: ./generate_mesh.sh
        - run: ./simulate --input mesh.exo
        - run: ./check_results output.csv

Save this as ktests.yml.

Running the Workflow

To run the workflow:

kuristo run -l /path/

Kuristo will traverse the directory structure and try to find ktests.yaml files with workflows. Then, it will execute each job in order, tracking progress and logging output into the .kuristo-out/ directory.

The command-line output will look like this:

✔ [1] simple test finished with return code 0 [1.01s]

✔ Success: 1    x Failed: 0    - Skipped: 0    Total: 1
  Took: 1.5s

By default, output is printed to the terminal and stored in per-run and per-job subdirectories under .kuristo-out/.

Options and Verbosity

You can control verbosity with the --verbosity option:

kuristo run workflow.yml --verbosity=2

Verbosity levels:

0: silent
1: errors only
2: default
3: detailed output for each step

List available jobs

Use this to see what jobs are would be executed:

kuristo list

This will traverse the directory structure from the current working directory and look for ktests.yaml files. You can specify different location via -l /path/to/start/search/from.

Environment diagnostics

Use the doctor command to generate a diagnostic report about your Kuristo environment:

kuristo doctor

This outputs detailed information including:

Kuristo version and Python interpreter
Platform and CPU configuration
Log and config file locations
MPI launcher
Active plugins, registered steps and actions
Logging and cleanup policies