Googled a character. Found out they die in a later season. Built Spol, an extension, so that never happens again
After having my show spoiled, I hypothesized that this may be a problem suited for the very LLMs trained on what is likely nearly every book, movie, show and transcript made to date. I would soon realize that, when provided with the right context, they do indeed have an incredible ability to sequence events.
The extension is a spoiler-free chat for media you’re currently watching. It tracks your progress and omits plot points from any further than where you have reached.
The goal is to make it safe to ask questions while watching without accidentally seeing future plot points from Google, wikis, Reddit threads, or autocomplete.
Right now, it works with Netflix and Crunchyroll. I’m planning to add more providers soon, and I’d appreciate suggestions for which ones people would want next.
Current limitations / work in progress:
- Support for querying arbitrary shows and progress points manually
- An embedded chat experience instead of a floating chat window
- More streaming providers
- Notices for media which is newer than model cutoff dates
There’s an always-free tier that is quite good for casual use. I also added a paid tier at $3/month for higher usage limits and a more capable model, which has been noticeably better at giving accurate answers while avoiding spoilers.
This is still in an early stage so there will be some bugs. Feedback is appreciated.