Run all tests except Enterprise Edition tests:

uv run pytest tests/ --ignore=tests/test_ee.py

Note: You can also disable EE-specific tools in the MCP server by adding this environment variable to your .env file:

KESTRA_MCP_DISABLED_TOOLS=ee

This prevents the MCP server from attempting to load any Enterprise Edition specific tools.

Run all tests including Enterprise Edition tests:

uv run pytest tests/

# Run only flow tests
uv run pytest tests/test_flow.py

# Run only execution tests
uv run pytest tests/test_execution.py

# Run multiple specific test files
uv run pytest tests/test_flow.py tests/test_execution.py

uv run pytest tests/ --ignore=tests/test_ee.py -v

To validate cross-version compatibility, run tests against all configured instances:

./tests/run_all_versions.sh        # default: --tb=short
./tests/run_all_versions.sh -v -x  # verbose, stop on first failure

The script loops through 4 Kestra instances (EE develop, OSS develop, EE latest, OSS latest), skips any that aren't reachable, and prints a summary at the end. Edit the INSTANCES array in the script to add/remove targets. Results are saved to .test-results/.

• test_ee.py - Enterprise Edition specific tests (requires EE/Cloud environment)
• test_*.py - All other tests work with OSS, EE, and Cloud editions

All MCP tools have integration test coverage except the 4 AI generation tools:

• generate_flow
• generate_dashboard
• generate_app
• generate_test

These tools delegate to Kestra's AI endpoints which depend on an external AI provider configuration. Their outputs are non-deterministic (LLM-generated YAML), making them unsuitable for automated assertions. They are best validated manually via the prompt files in docs/prompts/.