Perf/codecovtrigger by Josephrp · Pull Request #152 · DeepCritical/DeepCritical

Josephrp · 2025-10-13T09:59:21Z

Pull Request

Description

Type of Change

Component

Related Issues

Fixes #
Closes #
Related to #

Changes Made

Testing

I have tested these changes locally
I have added/updated tests for my changes
All existing tests pass
I have tested with different configurations
I have tested with different flows (PRIME, Bioinformatics, DeepSearch, etc.)

Test Configuration

# Example test command
uv run deepresearch question="..." app_mode=single_react

Configuration Changes

No configuration changes
Added new configuration options
Modified existing configuration
Removed configuration options

Configuration Details

# Example configuration changes
flows:
  new_flow:
    enabled: true
    params:
      new_param: "value"

Documentation

Performance Impact

No performance impact
Performance improvement
Performance regression (explain below)

Performance Details

Execution time:
Memory usage:
Other metrics:

Breaking Changes

No breaking changes
Breaking change (describe below)

Migration Guide

Checklist

My code follows the project's style guidelines
I have performed a self-review of my own code
I have commented my code, particularly in hard-to-understand areas
I have made corresponding changes to the documentation
My changes generate no new warnings
I have added tests that prove my fix is effective or that my feature works
New and existing unit tests pass locally with my changes
Any dependent changes have been merged and published

Additional Notes

Screenshots/Output

Before

After

Reviewer Notes

@defer

Removes all references to non-existent @defer decorator from codebase. The @defer decorator never existed in Pydantic AI. Tools are correctly implemented using standard Pydantic AI patterns. Changes: - Removed 16 @defer comments from tool files - Updated README Known Issues section - All tools continue to work correctly (no functional changes) Fixes #2

- Implement VLLMModel wrapper around existing VLLMClient - Add OpenAICompatibleModel for vLLM, llama.cpp, TGI servers - Provide factory methods (from_vllm, from_llamacpp, from_tgi) - Include streaming support and message conversion - Add convenience aliases for VLLMModel and LlamaCppModel

- Replace non-existent OpenAIProvider with OllamaProvider from pydantic_ai - Remove dataclass decorator to properly inherit from OpenAIChatModel - Fix factory methods to pass model_name as positional argument - Add comprehensive test suite with 8 passing tests - Skip integration tests that require actual vLLM servers

- Add from_config() method to support Hydra DictConfig - Update all factory methods (from_vllm, from_llamacpp, from_tgi, from_custom) to accept optional config - Support config override via direct parameters - Extract generation settings from config (temperature, max_tokens, etc.) - Add environment variable fallbacks (LLM_BASE_URL, LLM_API_KEY) - Create config files for llamacpp, tgi, and vllm - Update tests to cover both config-based and direct parameter approaches - All 10 tests passing

- Add LLMModelConfig and GenerationConfig datatypes - Remove redundant vllm_model.py - Update openai_compatible_model.py with validation - Rewrite tests to use actual config files (30 tests)

Integrated latest upstream changes while preserving Issue #10 LLM client work. Conflict resolutions: - DeepResearch/src/datatypes/__init__.py: Added LLM model types to __all__ without duplicates - Accepted upstream versions for bioinformatics_tools.py, deep_agent_tools.py, README.md

…ners

…in permissions Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com> Signed-off-by: Tonic <joseph@legml.ai>

…Critical into feat/addstools