NearAI LangChain Integration

nearai_langchain provides seamless integration between NearAI and LangChain, allowing developers to use NearAI's capabilities within their LangChain applications.

🚧 Development Status

This library is currently in active development.

🎯 Key Purposes

1. Optional NearAI Inference Integration

   • Access model inference through NearAI's optimized infrastructure
   • Maintain compatibility with standard LangChain for other use cases
   • Seamlessly switch between NearAI and LangChain inference

2. NearAI Registry Integration

   • Register and manage agents in the NearAI registry
   • Optionally, enable agent-to-agent interaction and make your agents callable by other agents in the NearAI registry
   • Auto-generate or validate agent metadata
   • Example metadata.json:

     {
       "category": "agent",
       "name": "example_agent",
       "namespace": "user.nearai",
       "tags": ["example"],
       "details": {
         "agent": {
           "defaults": {
             "model": "llama-v3p1-70b-instruct",
             "provider": "fireworks"
           },
           "framework": "langchain"  // Use "langchain" or "nearai" for inference
         }
       }
     }

3. Agent Intercommunication

   • Upload agents to be used by other agents
   • Call other agents from the registry in your own agents
   • Framework-agnostic: works with both NearAI and LangChain inference

4. Benchmarking and Evaluation

   • Run popular or user owned benchmarks on agents
   • Optionally, upload evaluation results to NearAI evaluation table
   • Support for both NearAI and LangChain inference frameworks

🌟 Features

• Drop-in replacement for LangChain chat models
• Support for multiple model providers
• Flexible framework switching between LangChain and NearAI
• Type-safe interfaces

🚀 Quick Start

import getpass
import os

import nearai_langchain
from langchain_core.messages import HumanMessage, SystemMessage

NearaiLangchain.init() # Reads metadata and inits either Langchain or NearAI. Supports "langchain" or "nearai" frameworks.

model = InferenceProvider()

messages = [
    SystemMessage("Translate the following from English into Italian"),
    HumanMessage("hi!"),
]

model.invoke(messages)

📦 Installation

pip install nearai-langchain

🛠️ Development Setup

1. Clone the repository:

   git clone https://github.com/nearai/nearai_langchain.git
   cd nearai_langchain

2. Install dependencies:

   ./install.sh

3. Development tools:

• Run format check: ./scripts/format_check.sh
• Run linting: ./scripts/lint_check.sh
• Run type check: ./scripts/type_check.sh