A Streamlit web app that transforms any goal into actionable, time-bound tasks using local open-source LLMs (Ollama) via the OpenAI Agents SDK. No API keys required — runs entirely on your machine.
Screenshot of the app in action — generating a task plan for renewing a dbt Certified Developer certificate.
-
Ollama installed and running. See ollama.com for setup.
-
A model pulled — e.g.
ollama pull llama3.2:3b
This project uses uv for package management. Dependencies are configured in pyproject.toml.
uv syncuv run streamlit run app.pyThe app will open at http://localhost:8501
Model: The app uses llama3.2:3b by default. To change, edit the model parameter in app.py:
model=LitellmModel(model="ollama_chat/llama3.2:3b"), # or ollama_chat/llama3.2, ollama_chat/mistral, etc.- "Learn a new programming language in 3 months"
- "Start contributing to open source and land a first PR"
- "Automate a recurring workflow (scripts, CI/CD, or Zapier/n8n)"
- "Set up a personal finance dashboard (track spending, savings, investments)"
- "Build a morning routine that includes exercise, reading, and planning"