Overview

azureaicommunity-agent-language-middleware is a plug-and-play translation layer for AI agent pipelines built on agent-framework. It automatically detects the user's language, translates incoming messages to the agent's target language, and back-translates the response — using Azure Translator with an optional LLM agent as fallback.

✨ Features

📦 Installation

pip install azureaicommunity-agent-language-middleware

🚀 Quick Start

import asyncio
from agent_framework import Agent
from agent_framework.ollama import OllamaChatClient
from language_middleware import LanguageTranslationMiddleware

async def main():
    client = OllamaChatClient(model="gemma3:4b")
    agent = Agent(client)

    middleware = LanguageTranslationMiddleware.create(
        llm_agent=agent,
        target_language="en",
    )

    try:
        result = await agent.run("Wie heißt die Hauptstadt von Frankreich?", middleware=middleware)
        print(result.text)  # "Die Hauptstadt von Frankreich ist Paris."
    finally:
        await middleware.aclose()

asyncio.run(main())

⚙️ Configuration

🧑‍💻 Usage

Option 1 — Azure Translator + LLM Fallback (Recommended)

import asyncio
import os
from agent_framework import Agent
from agent_framework.ollama import OllamaChatClient
from language_middleware import LanguageTranslationMiddleware, AzureTranslatorConfig

azure_config = AzureTranslatorConfig(
    key=os.environ.get("AZURE_TRANSLATOR_KEY", ""),
    region=os.environ.get("AZURE_TRANSLATOR_REGION", ""),
    endpoint=os.environ.get("AZURE_TRANSLATOR_ENDPOINT", ""),
)

async def main():
    client = OllamaChatClient(model="gemma3:4b")
    agent = Agent(client)

    middleware = LanguageTranslationMiddleware.create(
        azure_config=azure_config,
        target_language="en",
        min_confidence=0.8,
        llm_agent=agent,
    )

    try:
        tamil_query = "தஞ்சாவூரின் வரலாற்று சிறப்பு என்ன?"
        tamil_result = await agent.run(tamil_query, middleware=middleware)
        print("Tamil response:", tamil_result.text)

        german_query = "Was ist die historische Bedeutung von Thanjavur?"
        german_result = await agent.run(german_query, middleware=middleware)
        print("German response:", german_result.text)
    finally:
        await middleware.aclose()

asyncio.run(main())

Option 2 — LLM-Only (No Azure Subscription Required)

import asyncio
from agent_framework import Agent
from agent_framework.ollama import OllamaChatClient
from language_middleware import LanguageTranslationMiddleware

async def main():
    client = OllamaChatClient(model="gemma3:4b")
    agent = Agent(client)

    middleware = LanguageTranslationMiddleware.create(
        llm_agent=agent,
        target_language="en",
    )

    try:
        result = await agent.run("Wie heißt die Hauptstadt von Frankreich?", middleware=middleware)
        print(result.text)  # "Die Hauptstadt von Frankreich ist Paris."
    finally:
        await middleware.aclose()

asyncio.run(main())

Option 3 — Azure Only (No LLM Fallback)

middleware = LanguageTranslationMiddleware.create(
    azure_config=azure_config,
    target_language="en",
)

⚙️ How It Works

1. Intercept      →  middleware captures the last user message
2. Detect         →  language is detected with a confidence score
3. Translate      →  message is translated to the target language (if needed)
4. Agent runs     →  the primary LLM processes the translated message
5. Back-translate →  agent response is translated back to the user's original language
6. Return         →  user receives the response in their native language

Translation providers (with automatic fallback):

🤝 Contributing

Contributions are welcome! Please open an issue to discuss what you'd like to change before submitting a pull request.

1. Fork the repository
2. Create a feature branch (git checkout -b feature/my-feature)
3. Commit your changes (git commit -m 'Add my feature')
4. Push to the branch (git push origin feature/my-feature)
5. Open a Pull Request

📄 License

MIT — see LICENSE for details.