# Testing Best Practices
## Verify Before Assuming
When working with failing tests, always verify the actual behavior before assuming what the code does:
1. **Test actual behavior first**: Before changing tests to expect different behavior, run the actual code path to see what really happens.
```bash
# Example: test yq behavior directly before assuming limitations
uv run python -c "
from mcp_json_yaml_toml.yq_wrapper import execute_yq, FormatType
result = execute_yq('.key', input_file='test.toml', input_format=FormatType.TOML, output_format=FormatType.TOML)
print(f'stdout: {result.stdout!r}')
print(f'returncode: {result.returncode}')
"
```
2. **Don't assume tool limitations**: External tools like yq evolve. A limitation in an older version may be resolved in the current version. Check the DEFAULT_YQ_VERSION and test against it.
3. **Check version-specific capabilities**: When tests fail, verify whether the failure is:
- A bug in the test assertions
- A bug in the implementation
- An outdated assumption about tool capabilities
## Test Modification Guidelines
- If tests were previously passing in CI and now fail, the tests are likely correct
- Read test docstrings carefully - they explain what behavior is being tested
- When tests reference "improvements" or version-specific features, verify the current version supports them
