feat(exceptions): standardize provider error __str__ format

[lesbass]

Closes #113

Summary

Adds error_type class attribute and __str__ override to ProviderError and its subclasses, producing a consistent [provider] error_type: message format across all LLM providers.

Changes

• openagent_eval/exceptions/provider.py: Added error_type class attribute and __str__ override to ProviderError, ProviderNotFoundError, ProviderConnectionError, and ProviderExecutionError
• tests/unit/test_exceptions.py: Updated tests for new standardized format, added format tests with provider name

Format

Error class	error_type	Example output
ProviderError	"Provider Error"	[openai] Provider Error: API key not found
ProviderNotFoundError	"Provider Not Found"	[my_provider] Provider Not Found: Provider not found: my_provider
ProviderConnectionError	"Connection Error"	[gemini] Connection Error: Failed to connect
ProviderExecutionError	"Execution Error"	[anthropic] Execution Error: API call failed

When provider_name is None, the [...] prefix is omitted.

Verification

• Unit tests: 26/26 passed (test_exceptions.py)
• Ruff: clean on changed files
• Integration tests: not run (missing loguru dependency, pre-existing)

[himanshu231204]

Thanks for the PR! The CI tests are failing across all supported Python versions. Please check the workflow logs, fix the failures, and update the PR. 👍

[lesbass]

Thanks for the review! I found the issue — the __str__ override was dropping self.details from the output, which broke pytest.raises(match=...) patterns that match on detail keys (e.g., available_embedders, available_retrievers).

Fixed in the latest commit: the __str__ now includes non-provider_name details in the standardized format, matching the base OpenAgentEvalError behavior. The provider_name key is filtered out (already shown in the [provider] prefix).

[nope] Provider Not Found: Provider not found: nope (available_embedders=mock, sentence_transformers)
[gemini] Connection Error: Failed to connect (original_error=Connection failed)
[anthropic] Execution Error: API call failed (original_error=Execution failed)

All unit tests pass locally. CI should be green now.

[lesbass]

The __str__ fix has been pushed (commit 7e4510a). The issue was that __str__ wasn't including self.details, which caused pytest.raises(match=...) patterns matching on detail keys (available_embedders, available_retrievers) to fail.

Local verification (Python 3.13):

• All 26 exception tests pass
• The two previously failing tests (test_unknown_embedder, test_unknown_retriever) now pass
• All other tests that were passing before still pass

CI status: The latest CI run needs maintainer approval (first-time contributor policy). Could you approve the workflow? The previous run was at HEAD before the fix commit.

Pre-existing unrelated failures in CI (missing optional openai/tiktoken/pypdf dependencies and one flaky semantic similarity test) are unchanged by this PR.

[github-actions]

🎉 Congratulations @lesbass!

Your pull request has been successfully merged into main. 🚀

Thank you for contributing to OpenAgentHQ and helping improve the project.

We truly appreciate your contribution and hope to see you back with more amazing PRs!

Happy Open Sourcing! ❤️

🌟 This is your first merged contribution to this repository.
Welcome to the OpenAgentHQ contributors family!