This repository is a minimal, GitHub-ready template demonstrating an Agentic AI pattern:
- Base Agent (generates content)
- Reflection Agent (validates & improves)
- Pluggable model backend: Ollama (local) or OpenAI (cloud)
- Clone or download this repo.
- Install dependencies:
pip install -r requirements.txt - If using Ollama (recommended for local experimentation):
- Install and run Ollama server (see https://ollama.com)
- Pull models, e.g.:
ollama pull qwen2.5:14b-instruct ollama pull qwen3:7b-instruct
- Run:
python main.py
Edit config.py to switch backend or model names.
Default backend = Ollama.
models/model_client.py: abstracts model calls (Ollama/OpenAI)agents/base_agent.py: base test-case generatoragents/reflection_agent.py: reflection/repair agentmain.py: example runner and Excel export
- This template is intentionally minimal and educational — production systems need more error handling, authentication, rate-limiting, and observability.