Use classifier contract in identify flow

[henry3260]

Summary

• update LLMFingerprinter.identify() to consume classifier.classify_result() / MultiClassResult
• preserve the existing legacy dict shape consumed by the CLI report layer
• add focused tests for normal and OOD contract-to-report mapping

Tests

• .venv/bin/python -m pytest tests

Result: 29 passed, 1 existing datetime.utcnow() deprecation warning.

[Copilot]

Pull request overview

Refactors LLMFingerprinter.identify() to consume the new MultiClassResult contract from classifier.classify_result() while preserving the legacy report dict shape consumed by the CLI. Adds focused tests for the contract-to-report mapping covering both normal and OOD paths.

Changes:

• Replace predict_with_confidence() tuple consumption with classify_result() and map top_label, metadata['predicted_label'], metadata['is_ood'], metadata['probabilities'] (with ranking fallback), and metadata['ood_info'] to legacy keys.
• Update OOD/non-OOD log messages to use the new contract fields.
• Add tests/fingerprinter/test_identify_contract.py with a fake contract-only classifier asserting the legacy mapping and OOD best-guess preservation.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
llm_fingerprinter/fingerprinter.py	Rewrite identify() to consume MultiClassResult and map to legacy report keys
tests/fingerprinter/test_identify_contract.py	New tests verifying contract-to-legacy mapping for normal and OOD paths

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.