AI Log Analyzer

Serverless AI-powered log analysis tool using Claude API and AWS infrastructure.

Overview

This project automatically analyzes application logs using Claude AI (Anthropic) and provides:

• Summary of issues found
• Severity classification (critical/warning/info)
• Recommended actions

Built with AWS Lambda, API Gateway, and deployed via Terraform (Infrastructure as Code).

Architecture

User Request (HTTP POST)
         |
         v
    API Gateway
         |
         v
    AWS Lambda (Python 3.12)
         |
         v
    Claude API (Anthropic)
         |
         v
    JSON Response (AI Analysis)

Tech Stack

• Cloud: AWS (Lambda, API Gateway, IAM)
• AI: Claude Sonnet 4 (Anthropic)
• Language: Python 3.12
• IaC: Terraform
• Libraries: anthropic, boto3

Project Structure

ai-log-analyzer/
├── lambda/
│ ├── lambda_function.py # Main Lambda handler
│ └── (anthropic libraries)
├── terraform/
│ ├── main.tf # Infrastructure definition
│ ├── variables.tf # Configuration variables
│ ├── outputs.tf # API endpoint outputs
│ └── terraform.tfvars # Variable values (not in repo)
└── README.md

Quick Start

Prerequisites

• AWS Account with CLI configured
• Terraform >= 1.0
• Anthropic API key

Deployment

1. Clone repository

git clone https://github.com/marand85/ai-log-analyzer
cd ai-log-analyzer

2. Configure variables

cd terraform
cp terraform.tfvars.example terraform.tfvars

Edit terraform.tfvars with your Anthropic API key (use your preferred editor):

nano terraform.tfvars
# or: code terraform.tfvars
# or: vim terraform.tfvars

3. Deploy infrastructure

terraform init
terraform apply

4. Get API endpoint

terraform output api_endpoint

Usage

Test the API

curl -X POST https://your-api-endpoint/ai-log-analyzer \
  -H "Content-Type: application/json" \
  -d '{"logs": "ERROR 2025-01-16 12:00:00 Database connection timeout\nCRITICAL Payment processing failed"}'

Example Response

{
  "status": "success",
  "analysis": "## Log Analysis Report\n\n### Summary\n- Database timeout\n- Payment failure\n\n### Severity: CRITICAL\n\n### Recommended Actions\n- Check database connectivity\n- Review payment gateway status..."
}

⚠️ Clean Up (Important)

Always remember to destroy your infrastructure when you are done testing to avoid unexpected AWS charges:

terraform destroy

Key Features

• Serverless - No infrastructure management
• AI-Powered - Claude Sonnet 4 for intelligent analysis
• Infrastructure as Code - Full Terraform deployment
• Scalable - Auto-scaling Lambda + API Gateway
• Cost-Effective - Pay only for actual usage

Development

Local Testing

Package Lambda dependencies:

cd lambda
pip install anthropic -t .

Update code (use your preferred editor):

nano lambda_function.py
# or: code lambda_function.py
# or: vim lambda_function.py

Redeploy:

cd ../terraform
terraform apply

Security

• API key stored as encrypted environment variable
• IAM roles with least privilege access
• HTTPS-only API endpoint
• No authentication on API (add API Gateway authorizer for production)

Cost Estimate

Service	Usage	Cost
Lambda	1000 requests/month	~$0.20
API Gateway	1000 requests/month	~$0.01
Claude API	1000 requests	~$3.00
Total		~$3.21/month

Based on Free Tier + typical usage

Future Enhancements

• Add authentication (API Gateway Authorizer)
• Store analysis results in DynamoDB
• Add CloudWatch dashboards
• Support multiple AI models (GPT-4, etc.)
• Batch log processing

Author

Mariusz Andrzejewski AI Platform Engineer

• GitHub: https://github.com/marand85/ai-log-analyzer

License

MIT License