feat: enable label validation by default

[cmungall]

Summary

• Label validation (--labels) is now enabled by default for validate-data command
• This catches AI hallucinations where valid term IDs have incorrect labels - the primary use case for this validator
• Users can disable with --no-labels if they only want ID validation

Changes

• cli.py: Changed validate_labels default from False to True
• binding_plugin.py: Changed validate_labels parameter default from False to True
• Updated documentation (README.md, docs/binding-validation.md, docs/plugin-reference.md)

Test plan

• All unit tests pass (58 tests)
• Doctests pass
• mypy type checking passes
• ruff linting passes
• CLI help shows [default: labels]

Closes #8

🤖 Generated with Claude Code

[claude]

Claude finished @cmungall's task —— View job

---

PR Review: Enable Label Validation by Default

✅ Overall Assessment: EXCELLENT - This is a well-implemented, consistent change that improves the tool's primary use case.

🎯 Summary

This PR changes the default value of validate_labels from False to True for the validate-data command and BindingValidationPlugin. This enables label validation by default, which catches AI hallucinations where term IDs are valid but labels are incorrect.

✅ Strengths

1. Excellent Consistency

• ✅ CLI default changed in both validate_data() and validate_all() functions (cli.py:176, 409)
• ✅ Plugin constructor default changed (binding_plugin.py:95)
• ✅ Documentation updated across 3 files (README, binding-validation.md, plugin-reference.md)
• ✅ Docstring and example updated in plugin (binding_plugin.py:106, 16)
• ✅ Help text enhanced to indicate default behavior (cli.py:174)

2. Well-Motivated Change

• ✅ Addresses the primary use case: catching AI hallucinations
• ✅ Makes the "anti-hallucination guardrail" feature enabled by default
• ✅ Backward compatibility maintained via --no-labels flag
• ✅ Aligns tool behavior with its main value proposition

3. Clean Implementation

• ✅ Changes are minimal and surgical
• ✅ No breaking API changes - only default value shifts
• ✅ Clear commit message and PR description
• ✅ All tests reportedly pass

📝 Code Quality Assessment

Security: ✅ No Concerns

• No new security vectors introduced
• Changes only affect default validation behavior
• No data exposure or privilege escalation risks

Performance: ⚠️ Minor Impact

• Label validation adds ontology lookups (cached)
• Impact is positive for user experience vs. slight performance cost
• Existing caching mechanisms mitigate performance concerns
• Users can disable with --no-labels if needed

Maintainability: ✅ Excellent

• Changes follow established patterns
• Documentation thoroughly updated
• No technical debt introduced

🧪 Test Coverage Observations

The test suite appears comprehensive:

• ✅ 58 tests pass (as reported in PR)
• ✅ Unit tests explicitly set validate_labels=True/False as needed
• ✅ CLI tests use --cache-dir showing awareness of the feature
• ✅ Integration tests cover the validation workflow

Note: The test files in tests/test_plugins.py explicitly set validate_labels=False in many tests, which means they're testing the specific behavior they need rather than relying on defaults. This is good practice and shows the change won't break existing tests.

🚀 Recommendations

This PR: Ready to Merge

1. ✅ Approve and merge - this is a solid improvement
2. Consider adding a note to CHANGELOG.md about the default change
3. The change is well-documented and backward-compatible

Future Enhancements (not for this PR)

1. Consider adding CLI examples in help text showing both --labels and --no-labels
2. Could add metrics/logging for label validation hit rates to measure impact

🎯 Decision Rationale

This change makes perfect sense because:

• Primary use case: The tool's main value is catching hallucinated ontology terms
• User experience: Users expect anti-hallucination features to be enabled by default
• Safe change: Backward compatibility via explicit --no-labels flag
• Well-executed: Comprehensive implementation across codebase

✅ Final Verdict: LGTM

This is exactly the type of thoughtful, user-centric improvement that enhances a tool's usability. The implementation is thorough, consistent, and well-documented.

Recommendation: Approve and merge immediately.

---

[Copilot]

Pull request overview

This PR changes the default behavior of label validation from opt-in to opt-out by setting validate_labels=True as the default for both the CLI and the BindingValidationPlugin. This catches AI hallucinations where valid term IDs have incorrect labels, addressing the primary use case for the validator. Users who only want ID validation can use the --no-labels flag to disable it.

Key Changes

• Default value changed from False to True for the validate_labels parameter in both the CLI and plugin
• Updated documentation across README.md and docs/ to reflect the new default behavior
• Updated doctest example to show the new default value

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
src/linkml_term_validator/plugins/binding_plugin.py	Changed validate_labels parameter default from False to True and updated docstring
src/linkml_term_validator/cli.py	Changed validate_labels default from False to True in both validate_data and validate_all commands
docs/plugin-reference.md	Updated documentation to reflect new default value and clarified that it's enabled by default
docs/binding-validation.md	Updated parameter table to show True as the default
README.md	Updated example configuration comment to show default: true

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated no new comments.