Foreseer - a mostly AI-assisted personal Seerr fork with SuggestArr-style discovery
I made a personal fork of Seerr called Foreseer:
https://github.com/selmant/seerr
The short version: I wanted Seerr to have more of the discovery and recommendation workflow that SuggestArr provides, without needing to run a separate companion service.
I’m not planning to add AI-powered recommendations. I tried them, but they weren’t useful for my workflow, so Foreseer uses Trakt for suggestions instead.
Originally, I was using SuggestArr and started forking it to add features that made it feel more like Seerr. I was even using Seerr backends in that fork. Eventually, I decided it made more sense to fork Seerr itself and add the SuggestArr-style functionality there.
Foreseer still keeps Seerr’s core request workflow and integrations with Plex, Jellyfin, Emby, Radarr, and Sonarr. So far I’ve added:
- Trakt recommendations, lists, watchlists, and browsing
- Trakt OAuth account linking
- Trakt watched/unwatched history integration
- Trakt watch and rating actions
- Multi-source rating badges, including IMDb, Rotten Tomatoes, Metacritic, Trakt, and TMDB
- Discover filters for quality, anime routing, watched content, and Trakt history
- Instant movie requests and season-one TV requests
- Better rating hover interactions and rating popovers
- Seerr-compatible config paths and migration behavior
Full disclosure: this project is mostly AI-assisted. I use AI heavily for implementation, refactoring, and documentation, while testing it against my own setup and fixing the problems I encounter.
This is primarily for my own use. Development will be driven by my personal needs, and I’m not committing to a public roadmap, feature requests, or PR reviews. Bug-fix PRs are welcome, though.
I understand that AI-assisted projects aren’t for everyone, but I’m not looking to debate that aspect. It’s a personal fork built for my own setup, so if it isn’t useful to you, feel free to skip it.
If you do try it, bug reports with reproduction steps are welcome. It’s still early-stage software, so expect rough edges and possible breaking changes.