Can I Watch? A helpful little tool for a scary movie night.

Pascal Wiltschko — Sun, 18 Oct 2026 20:38:21 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle*

What I Built

Love horror, hate surprise animal cruelty? Same movie, very different evening.

I built caniwatch, a Python CLI that checks a movie against the things you'd rather avoid. Give it a title and your concerns, and it researches content guides, shows its sources, and offers a short recommendation.

Missing information stays "unknown." No guessing that silence means you're good to go.

The potential paying audience is movie fans who want personal content advice without sharing sensitive concerns with a hosted AI service. A small paid app could make this useful beyond the terminal. For now, it's a working CLI prototype, built solo.

Demo

Code

Source code on GitHub

The code is MIT licensed, with setup instructions in the README.

How I Built It

I used Python, Typer, Rich, and Pydantic, with Gemma 4 running locally through Ollama.

TMDB identifies the movie, and Tavily finds general content advisories. The local model then compares that evidence against your concerns and produces a structured assessment with source references.

Your concerns never shape the public search queries. They stay on your machine.

Why Does Open Innovation Matter?

Personal content concerns can be, well, personal.

Running an open-weight model locally lets me offer that assessment without sending those details to a hosted model API. I can inspect the prompts, change the model, and improve the workflow myself.

The movie research still needs internet access. The personal part runs locally, without a hosted inference bill on every check. That gives a small project like this more room to become a viable product.

My First User

My partner! She didn't pay but liked the concept, and we plan to use it for future scary movie evenings.

Happy Halloween, everyone!