Agentic Reasoning: Reasoning LLM with Agentic Tools

An open-source framework for deep research and beyond. The core idea is to integrate agentic tools into LLM reasoning.

Warning: Still in development. Theoretically runnable, but undergoing rapid updates.

Install

install from environment.yml, e.g.

conda env create -f environment.yml

Run

export your remote LLM API key in environment if you are using remote LLM, e.g.

export OPENAI_API_KEY="your openai api key"

export your you.com api key in environment if you are using deep research

export YDC_API_KEY="your you.com api key"

prepare JINA and BING API key if needed

run command:

python scripts/run_agentic_reason.py \
--use_jina True \
--jina_api_key "your jina api key" \
--bing_subscription_key "your bing api key"\ 
--remote_model "your remote model name, e.g. gpt-4o" \
--mind_map True \ (optional)
--deep_research True \ (optional, if you want to use deep research)

TODO LIST

• auto research
• clean agentic reasoning

Thanks

Code copied a lot from ...

Ref

@misc{wu2025agenticreasoningreasoningllms,
      title={Agentic Reasoning: Reasoning LLMs with Tools for the Deep Research}, 
      author={Junde Wu and Jiayuan Zhu and Yuyuan Liu},
      year={2025},
      eprint={2502.04644},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2502.04644}, 
}