chat-tools 2.2

Tools for human-machine interaction based on large language models.

chattools

The most simple tool for AI chat, such as gemini, deepseek, ollama

plz save the API keys in .env.key file in the current path.

Example

run python path/to/test.py (deepseek)

run python path/to/test-mistral.py to utilize mistral AI.

Make CLI

# default model is `gpt-oss:120b`

from chattools import OllamaChat

description="Intelligent enough to help me for anything."
name="Asistant"

with OllamaChat(description=description, name=name) as chat:
    chat.run()

if __name__ == "__main__":
    from fire import Fire
    Fire()

Code

# import YourLLM API 
from mixin import ChatMixin
from utils import get_api_key

# api_key = get_api_key


class YourChat(ChatMixin, YourLLM):

    def __init__(self, description=None, history=[], name='Assistant', model="model-name", *args, **kwargs):
        super().__init__(api_key=api_key, *args, **kwargs)  # init method of super class
        self.description = description
        self.name = name
        self.model = model
        self.chat_params = {}

        self.history = history

    def _reply(self, messages, max_retries=100):
        """The wrapper method of the original `chat` method
        """

        k = 0
        while True:
            # try `max_retries` times
            try:
                """
                get the response of the model. such as 
                self.chat.completions.create(
                        model=self.model,
                        messages=messages,
                        **self.chat_params)
                """
            except:
                ...

Commands

register a command as follows

from chat_tools.commands import Commands

@Commands.register("read")
def read_history(obj, path):
    # obj.history = read from `path`
    pass

use the command in the chat as !read path

---