open-taranis 0.1.4

Minimalist Python framework for AI agents logic-only coding with streaming, tool calls, and multi-LLM provider support

open-taranis

Python framework for AI agents logic-only coding with streaming, tool calls, and multi-LLM provider support.

Installation

pip install open-taranis --upgrade

Quick Start

import open_taranis as T

client = T.clients.openrouter("api_key")

messages = [
    T.create_user_prompt("Tell me about yourself")
]

stream = T.clients.openrouter_request(
    client=client,
    messages=messages,
    model="mistralai/mistral-7b-instruct:free", 
)

print("assistant : ",end="")
for token, tool, tool_bool in T.handle_streaming(stream) : 
    if token :
        print(token, end="")

To create a simple display using gradio as backend :

import open_taranis as T
import open_taranis.web_front as W
import gradio as gr

gr.ChatInterface(
    fn=W.chat_fn_gradio(
    client=T.clients.openrouter(API_KEY),
    request=T.clients.openrouter_request,
    model="mistralai/mistral-7b-instruct:free",
    _system_prompt="You are an agent named **Taranis**"
).create_fn(),
    title="web front"
).launch()

Documentation :

• Base of the docs (coding some things before the real docs)

Roadmap

• v0.0.1: start
• v0.0.x: Add and confirm other API providers (in the cloud, not locally)
• v0.1.x: Functionality verifications in examples
• > v0.2.0: Add features for logic-only coding approach
• v0.6.x: Add llama.cpp as backend in addition to APIs
• v0.7.x: Add reverse proxy + server to create a dedicated full relay/backend (like OpenRouter), framework usable as server and client
• v0.8.x: Add PyTorch as backend with transformers to deploy a remote server
• v0.9.x: Total reduction of dependencies for built-in functions (unless counter-optimizations)
• v1.0.0: First complete version in Python without dependencies

Changelog

• v0.0.4 : Add xai and groq provider
• v0.0.6 : Add huggingface provider and args for clients.veniceai_request
• v0.1.0 : Start the docs, add update-checker and preparing for the continuation of the project...
• v0.1.1 : Code to deploy a frontend with gradio added (no complex logic at the moment, ex: tool_calls)
• v0.1.2 : Fixed a display bug in the web_front and experimentally added ollama as a backend
• v0.1.3 : Fixed the memory reset in the web_front and remove ollama module for openai front (work 100 times better)
• v0.1.4 : Fixed web_front for native use on huggingface, as well as handle_streaming which had tool retrieval issues

Advanced Examples

• tools call in a JSON database
• tools call in a HR JSON database in multi-rounds

Links

• PyPI
• GitHub Repository