https://github.com/persteenolsen/fastapi-jwt-auth-ai-agent-one
Python FastAPI with JWT Auth serving a LangChain ReAct AI Agent running at Vercel
https://github.com/persteenolsen/fastapi-jwt-auth-ai-agent-one
ai ai-agent fastapi jwt langchain python
Last synced: 12 days ago
JSON representation
Python FastAPI with JWT Auth serving a LangChain ReAct AI Agent running at Vercel
- Host: GitHub
- URL: https://github.com/persteenolsen/fastapi-jwt-auth-ai-agent-one
- Owner: persteenolsen
- Created: 2026-06-07T06:26:34.000Z (22 days ago)
- Default Branch: main
- Last Pushed: 2026-06-08T16:26:02.000Z (21 days ago)
- Last Synced: 2026-06-08T17:30:04.220Z (21 days ago)
- Topics: ai, ai-agent, fastapi, jwt, langchain, python
- Language: Python
- Homepage: https://fastapi-jwt-auth-ai-agent-one.vercel.app/docs
- Size: 22.5 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# 🚀 FastAPI with JWT Auth serving a LangChain ReAct AI Agent
A production-style AI Agent API built with **FastAPI**, featuring **JWT authentication**, **Groq-powered LLMs**, and a **LangChain ReAct agent** with tool usage (Wikipedia).
This project demonstrates how to build a tool-using AI system that can:
- 🧠 Reason about user intent
- 🌐 Use external tools when needed
- 💬 Respond directly for conversational and creative tasks
---
## 📌 Project Info
- 📦 Version: 0.0.2
- 🐍 Python: 3.12
- 📅 Last Updated: 10-06-2026
---
## ✨ Features
### 🔐 Authentication (JWT)
- Secure login system using JWT (HS256)
- Protected endpoints with Bearer token
- Token expiration support
- Environment-based credentials (.env)
---
### 🤖 AI Agent (LangChain ReAct)
- 🧠 Reasoning + Acting loop (ReAct pattern)
- 🔁 Multi-step decision making
- 🎯 Smart tool selection (Wikipedia only when needed)
- 🛑 Controlled iterations (prevents infinite loops)
- 🧯 Parsing error handling for stability
---
### 🧠 LLM Integration (Groq)
- ⚡ Model: `llama-3.3-70b-versatile`
- 🚀 Ultra-fast inference via Groq API
- 🎛️ Deterministic outputs (`temperature = 0`)
---
### 🌐 Wikipedia Tool
- 📚 Wikipedia API integration via custom wrapper
- 🔎 Factual knowledge retrieval
- 🔁 Retry support for transient API failures
- 🧯 Graceful fallback on errors
- 🧠 Used only for factual queries
---
### 🧩 Agent Capabilities
- ❓ General question answering
- 📚 Factual lookup (Wikipedia)
- 💬 Conversational chat
- 😂 Joke generation
- ✍️ Creative writing (poems, stories)
- 🧠 Tool-augmented reasoning
---
## 📡 API Endpoints
| Method | Endpoint | Description |
|--------|----------|-------------|
| 🔐 POST | `/login` | Get JWT token |
| 💬 POST | `/chat` | Chat with AI agent (protected) |
| ❤️ GET | `/health` | Service health check |
| 🧠 GET | `/test-groq` | Test LLM connection |
| 🌐 GET | `/test-wikipedia` | Test Wikipedia tool |
---
## ⚙️ Getting Started
### 1️⃣ Clone Repository
Clone the project from GitHub and enter the project directory.
---
### 2️⃣ Create Virtual Environment
Windows:
venv\Scripts\activate
Linux / macOS:
source venv/bin/activate
---
### 3️⃣ Install Dependencies
pip install -r requirements.txt
---
## 🔑 Environment Variables
Create a `.env` file:
SECRET_KEY=your_secret_key_here
GROQ_API_KEY=your_groq_api_key
FAKE_USERNAME=admin
FAKE_PASSWORD=password
Generate a secure key:
python -c "import secrets; print(secrets.token_hex(32))"
---
## ▶️ Run the Application
uvicorn main:app --reload
App:
http://127.0.0.1:8000
Docs:
http://127.0.0.1:8000/docs
---
## 🔐 Authentication Flow
1. Call `/login`
2. Receive JWT token
3. Use token in requests:
Authorization: Bearer YOUR_TOKEN
---
## 🧠 How the Agent Works
User Input
↓
Agent Reasoning
↓
Wikipedia Tool (if needed)
↓
Direct Response (jokes, chat, creative writing)
↓
Final Answer
---
## 🏗️ Architecture
LLM:
Groq - llama-3.3-70b-versatile
Agent:
LangChain ReAct Agent
Tools:
Wikipedia API
Execution Settings:
Max iterations: 2
Parsing error handling enabled
Verbose logging enabled
---
## 💬 Example Requests
### 😂 Joke Request
{
"message": "Tell me a joke"
}
Response:
{
"response": "Here's one: Why couldn't the bicycle stand up by itself? Because it was two-tired."
}
---
### 📚 Factual Question
{
"message": "What is the capital of France?"
}
Response:
{
"response": "Paris"
}
---
### ✍️ Creative Request
{
"message": "Write a short poem"
}
Response:
{
"response": "The sun sets slow and paints the sky..."
}
---
### 🌐 Wikipedia Tool Usage
{
"message": "Who is Albert Einstein?"
}
Response:
{
"response": "Albert Einstein was a theoretical physicist known for the theory of relativity."
}
---
## 📌 Use Cases
- 🤖 AI assistants
- 🎓 Educational tools
- 🧪 LangChain experimentation
- 🧠 Tool-augmented LLM systems
- ⚡ FastAPI backend AI services
---
## 🚧 Limitations
- 📚 Only Wikipedia as external tool
- 🧠 No long-term memory
- 🔐 Demo authentication system
- 🌐 Limited external knowledge sources
---
## 🚀 Future Improvements
- 🧠 Conversation memory
- 🔌 More tools (search, APIs, DB integration)
- 👥 Multi-user system
- 🔄 Token refresh system
- 🧩 Modular tool registry
- 📊 Analytics dashboard
---
## 📄 License
MIT License
---
## 🙌 Summary
This project combines:
⚡ FastAPI
🔐 JWT Authentication
🧠 Groq LLM (Llama 3.3 70B)
🤖 LangChain ReAct Agent
🌐 Wikipedia Tool
to build a **working tool-augmented AI system** that can reason, act, and respond intelligently depending on user intent.