chatloop 0.1.0

A reusable agentic chat library with tool-calling loop, multi-provider LLM support, and MCP integration.

chatloop

A reusable agentic chat library with tool-calling loop, multi-provider LLM support, and MCP integration.

Install

pip install chatloop[openai]          # OpenAI provider
pip install chatloop[anthropic]       # Anthropic provider
pip install chatloop[google]          # Google Generative AI provider
pip install chatloop[all]             # All providers + MCP
pip install chatloop[fastapi]         # FastAPI integration

Quick Start

import asyncio
from chatloop import ChatLoop, tool

@tool(description="Add two numbers")
async def add(context, a: int, b: int):
    return {"result": a + b}, None

async def main():
    loop = ChatLoop(
        provider="openai",
        model="gpt-4o",
        api_key="sk-...",
        tools=[add],
    )
    response = await loop.ask("What is 2 + 3?")
    print(response.answer)

asyncio.run(main())

Multi-Provider

# OpenAI
loop = ChatLoop(provider="openai", model="gpt-4o", api_key="...")

# Anthropic
loop = ChatLoop(provider="anthropic", model="claude-sonnet-4-5-20250514", api_key="...")

# Google
loop = ChatLoop(provider="google", model="gemini-2.0-flash", api_key="...")

MCP Servers

from chatloop import ChatLoop
from chatloop.mcp import MCPServer

postgres = MCPServer(
    command="npx", args=["-y", "@modelcontextprotocol/server-postgres", DB_URL],
    name="postgres",
)

async with ChatLoop(
    provider="openai", model="gpt-4o", api_key="...",
    mcp_servers=[postgres],
) as loop:
    response = await loop.ask("What tables exist in the database?")

FastAPI Integration

from fastapi import FastAPI
from chatloop import ChatLoop
from chatloop.integrations.fastapi import create_router

app = FastAPI()
loop = ChatLoop(provider="openai", model="gpt-4o", api_key="...", tools=[...])
app.include_router(create_router(loop=loop, prefix="/chat"))

License

MIT