IR-Elasticsearch-Agent

Quick project using LangChain create_agent method to build an agent with memory for Information Retrieval. The agent can answer questions using a retrieval tool and can also index documents on demand.

Purpose

Elasticsearch + LangChain agent with conversational memory for Information Retrieval tasks.

Setup

Install dependencies

uv sync

Start Elasticsearch and Kibana

docker-compose up -d

Elasticsearch will be available at http://localhost:9200
Kibana will be available at http://localhost:5601

Configure environment

Create a .env file:

ANTHROPIC_API_KEY=your_anthropic_api_key
EMBEDDINGS_MODEL_NAME =your_embedding_model_name
LLM_MODEL_NAME =your_llm_model_name
ES_HOST=your_elasticsearch_host

Usage

uv run main.py

Ask questions like:

• "Find articles about machine learning"
• "Search for technology articles"
• "What is deep learning?"

Type exit or quit to quit.