🤖 AI Agent for Application Log Analysis

AI-powered agent that analyzes application logs in real time, detects anomalies, notifies users, suggests fixes, and helps solve problems. Supports both local (via Ollama) and API-based LLMs. Includes an interactive web interface and provides integration with monitoring tools to streamline issue diagnosis and resolution.

---

📌 Key Features

• Real-time log analysis powered by LLMs (OpenAI, Anthropic, DeepSeek, Ollama, etc.)
• Interactive chat UI (TypeScript/HTML frontend)
• REST API via FastAPI
• Persistent chat history in PostgreSQL
• Log validation using Pydantic & PydanticAI
• Asynchronous backend (FastAPI, asyncpg)
• Observability with Logfire
• Short-term log storage in Redis
• Grafana/Prometheus monitoring stack (optional)

---

📦 Project Structure

BACKEND/
├── main.py                      # FastAPI app entry point
├── schemas.py                   # Pydantic models for chat/logs
├── utilslib.py                  # Log parsing, validation, and utility functions
├── pyproject.toml               # Python project config
├── uv.lock                      # uv dependency lock file
├── sample.env                   # Example environment config
├── README.md                    # This file
├── LLM_Agents/
│   └── agentslib.py             # LLM agent logic, system prompts, tools
├── Mock_UI/
│   ├── chat_app.html            # HTML frontend
│   ├── chat_app.ts              # TypeScript frontend logic
│   └── styles.css               # UI styles
├── Mock_Services/
│   └── sent_logs.ipynb          # Notebook for mock log sending
├── Postgres_DB/
│   ├── DB_PG17.py               # Async PostgreSQL logic
│   └── initdb17/
│       ├── docker-compose.yml   # Docker Compose for PostgreSQL
│       └── init_db.sql          # DB initialization script
├── Redis_DB/
│   └── ST_DB_Redis.py           # Async Redis logic for log storage
├── grafana/
│   ├── docker-compose.yml       # Docker Compose for Grafana/Prometheus
│   ├── prometheus.yml           # Prometheus config
│   ├── node_exporter/           # Node exporter for metrics
│   ├── LICENSE
│   └── NOTICE
├── static/
│   └── styles.css               # Additional static styles
├── test_logs/
│   ├── deanonymized_server.log
│   └── deanonymized_server_backup.log

---

🚀 How to Run

1. Clone the repository

   git clone https://github.com/el-arma/AGH_diploma
   cd AGH_diploma

2. Set up Python environment

   uv venv
   .venv\Scripts\activate
   uv sync

3. Set up PostgreSQL

   cd Postgres_DB\initdb17
   docker compose up -d

4. Set up Redis (short-term log storage)

   docker run -d --name redis-stack -p 6379:6379 redis/redis-stack-server:latest

5. Configure environment variables

   • Copy sample.env to .env and fill in your API keys (e.g., OPENAI_API_KEY, ANTHROPIC_API_KEY).

6. (Optional) Run Grafana/Prometheus monitoring

   cd grafana
   docker compose up -d

7. Run the application

   uvicorn main:app --host 127.0.0.1 --port 8000 --reload

   Open your browser: http://127.0.0.1:8000

---

🔌 API Endpoints:

External (for users)

Method	Endpoint	Description
POST	/logs/ingest	Submit logs for async LLM analysis

Internal (for internal system)

Method	Endpoint	Description
GET	/	Redirect to UI (main chat page)
GET	/chat_app.ts	Download TypeScript frontend logic
GET	/chat/	Retrieve chat history (main endpoint for frontend)
POST	/chat/	Send message to chat (streams LLM replies)
DELETE	/chat/delete	Delete chat(s) by chatId
POST	/set_model/	Change LLM model (OpenAI, Anthropic, DeepSeek, Ollama)

Explanation:

• External – endpoints for end users
• Internal – endpoints for system

---

⚙️ Tech Stack

• FastAPI – Async Python web framework
• PydanticAI – AI Agent Framework
• PostgreSQL 17 – Main DB
• asyncpg – Fast PostgreSQL driver
• Pydantic – Data validation & parsing
• Logfire – Observability and tracing
• OpenAI / Anthropic / DeepSeek / Ollama – LLM integrations
• Redis – Short-term log storage
• Grafana/Prometheus – Monitoring stack

---

🧩 Customization

• Change LLM model in LLM_Agents/agentslib.py
• DB settings in Postgres_DB/DB_PG17.py
• Extend schemas in schemas.py to match your data structures

---

🥊 Development Notes

• All DB operations are fully asynchronous
• Log analysis is handled as background tasks
• Logfire spans help trace DB and AI Agent actions

---