🤖 Project Architect Agent

A sophisticated AI agent designed to help architect, plan, and research software projects. It combines the power of Large Language Models (LLMs) with real-time web research capabilities and a developer-centric terminal interface.

✨ Features

• 🧠 Intelligent Research: Leverages the Tavily API to perform deep web searches and extract relevant content to answer complex architectural queries.
• 💬 Conversational Interface: A robust Terminal User Interface (TUI) built with Textual, offering a smooth, chat-like experience directly in your terminal.
• 🏗️ Specialized Knowledge: Equipped with a custom system prompt designed for high-level software architecture and project planning advice.
• 🔍 Observable: Fully integrated with LangSmith for tracing, debugging, and monitoring agent performance.
• 💾 State Persistence: Uses InMemorySaver to maintain conversation context within a session.

🛠️ Tech Stack

• AI Core: LangChain & LangGraph
• LLM: OpenAI GPT-4
• Tools: Tavily Search API
• UI: Textual (Python TUI)
• Observability: LangSmith

📐 Architecture

The agent is built as a graph-based application using LangGraph:

1. Input: User queries are captured via the Textual TUI.
2. Processing: The LangChain agent (OpenAI GPT-4) analyzes the request.
3. Tool Usage: If research is needed, the agent calls tavily_search or tavily_extract.
4. Response: The agent synthesizes findings and generates a markdown-formatted response.
5. Output: The response is streamed back to the TUI in real-time.

🚀 Prerequisites

• Python 3.13+
• Docker (optional, for containerized execution)

🔑 Environment Variables

Create a .env file in the root directory with the following keys:

Variable	Description
OPENAI_API_KEY	Required for the LLM (GPT-4).
TAVILY_API_KEY	Required for web search and extraction capabilities.
LANGCHAIN_API_KEY	Required for LangSmith observability.
LANGCHAIN_TRACING_V2	Set to true to enable LangSmith tracing.
LANGCHAIN_PROJECT	(Optional) Project name for LangSmith.

📦 Setup & Usage

Local Execution

1. Clone the repository:

   git clone https://github.com/andrecodea/project-architect-agent.git
   cd project-architect-agent

2. Install dependencies:

   pip install .

3. Run the agent:

   python agent.py

Docker Execution

1. Build and run with Docker Compose:

   docker compose up

   Note: The agent runs in interactive mode, so you must have a TTY attached.

2. To stop: Press Ctrl+C.