Radarr Movie Recommender

A local AI companion for Radarr : finds films you'll actually want to watch, completes your collections, and analyzes your taste. No API key, no cloud, no subscription. Just Ollama running on your machine.

The simplest use case — set it and forget it

Run it once manually, pick the films you want, or just let it add everything automatically:

python newmovies.py          # review and choose
python newmovies.py --auto   # add top 10 recommendations silently

Set it as a scheduled task and wake up every morning with new films already added to Radarr:

Windows (Task Scheduler):

$action = New-ScheduledTaskAction -Execute "python" -Argument "C:\path\to\newmovies.py --auto"
$trigger = New-ScheduledTaskTrigger -Daily -At "03:00"
Register-ScheduledTask -TaskName "RadarrRecommender" -Action $action -Trigger $trigger -RunLevel Highest

Linux/Mac (cron):

0 3 * * * cd /path/to/radarr-movie-recommender && python newmovies.py --auto

Your collection grows by itself , every morning if you want, 10 new films picked from your library's taste profile, validated against IMDb, added to Radarr and ready to download.

---

Previews

Classic mode — recommendations based on your library:

Saga mode — automatically complete a franchise:

Mood mode — find films by atmosphere:

---

What you can do that no other tool offers

🎭 Describe what you're in the mood for — in plain English

python newmovies.py --mood "dark and intense psychological thriller"
python newmovies.py --mood "samurai and honor in feudal japan"
python newmovies.py --mood "noir detective in a rainy city"
python newmovies.py --mood "feel good sunday afternoon comedy"
python newmovies.py --mood "mind-bending sci-fi with a twist ending"
python newmovies.py --mood "heist with a brilliant plan" --imdb-min 7.5

🎬 Start from any film — even one you don't own

python newmovies.py --like "Parasite"
python newmovies.py --like "2001: A Space Odyssey" --mood "existential and slow burn"
python newmovies.py --like "Inception" --mood "mind-bending"

🎞️ Complete an entire franchise automatically

python newmovies.py --saga "Star Wars"          # finds every missing episode
python newmovies.py --saga "Planet of the Apes" # original + reboot series
python newmovies.py --saga                      # auto-detects ALL incomplete sagas

🎥 Explore complete filmographies

python newmovies.py --director "Stanley Kubrick"
python newmovies.py --actor "Al Pacino"
python newmovies.py --composer "Ennio Morricone" --export morricone.html
python newmovies.py --author "Cormac McCarthy"   # all film adaptations

👥 Multi-actor search — unique to this tool

# Films featuring ANY of these actors
python newmovies.py --actor "Ben Stiller, Owen Wilson"

# Films where ALL of them appear TOGETHER
python newmovies.py --cast "Ben Stiller, Owen Wilson"
# → Zoolander, Starsky & Hutch, Night at the Museum...

python newmovies.py --cast "Robert De Niro, Al Pacino"
# → The Godfather Part II, Heat, Righteous Kill

🧠 AI analysis of your collection

python newmovies.py --analyze
# → "Your collection excels at 90s drama but lacks Kurosawa, French New Wave..."
# → Suggests 10 films to fill the gaps

python newmovies.py --stats
# → Genre breakdown, decade distribution, average rating

python newmovies.py --watchlist letterboxd_watchlist.csv
# → Import directly from your Letterboxd or IMDb watchlist

📋 Review with full plot synopsis

python newmovies.py --actor "Al Pacino" --synopsis
# Shows full plot before you decide to add each film:
#   + Serpico (1973)  IMDb:7.7
#   │ An honest New York cop named Frank Serpico blows the whistle
#     on rampant corruption in the police department.
#   add? (y/n):

📤 Export to HTML or CSV

python newmovies.py --actor "Ennio Morricone" --export morricone.html
python newmovies.py --analyze --export gaps.csv
# → Beautiful dark-themed HTML report you can share

---

Actual results

Command	What it found
--mood "dark and intense"	No Country for Old Men, Martyrs, Let the Right One In
--mood "samurai feudal japan"	Seven Samurai, Rashomon, Yojimbo, Throne of Blood, Samurai Rebellion
--mood "noir detective rainy city"	Se7en, Double Indemnity, Chinatown, Maltese Falcon, Big Sleep
--like "2001: A Space Odyssey"	Solaris, Stalker, Silent Running, Moon, Arrival
--saga "Star Wars"	All 10 missing episodes + Rogue One + Solo
--saga "Planet of the Apes"	6 missing films across original + reboot series
--director "Stanley Kubrick"	Full filmography, only missing titles
--actor "Al Pacino"	Serpico, Godfather I & II, Dog Day Afternoon, Scarface...
--composer "Ennio Morricone"	19 missing films incl. GBU, Once Upon a Time in America
--cast "De Niro, Pacino"	Exactly 3 films where both appear: Godfather II, Heat, Righteous Kill
--analyze	Detected gaps in Kurosawa, Bergman, French New Wave → 10 suggestions
--watchlist	Imports Letterboxd/IMDb CSV directly into Radarr

---

How it works

Your Radarr library
        │
        ▼
  Ollama (local LLM) understands theme, tone, atmosphere
        │
        ▼
  OMDb validates each suggestion
  (rating, year, genre, not already owned)
        │
        ▼
  Scoring: genre + director + cast + plot embeddings
        │
        ▼
  Results added to Radarr — with your approval

---

Requirements

• Python 3.10+
• Ollama running locally
• A running Radarr instance
• A free OMDb API key (1000 req/day)

---

Installation

git clone https://github.com/nikodindon/radarr-movie-recommender.git
cd radarr-movie-recommender
pip install -r requirements.txt
ollama pull llama3.1:8b
cp config.yaml.example config.yaml   # edit with your settings

config.yaml:

omdb_keys: your_key1,your_key2
radarr_api_key: your_radarr_api_key
radarr_url: http://localhost:7878/api/v3
root_folder: "D:\\Movies"
ollama_model: llama3.1:8b
quality_profile_id: 6
minimum_availability: announced

---

All commands

# Classic — based on your library
python newmovies.py
python newmovies.py --auto           # add everything without prompting
python newmovies.py --genre "Horror"

# Mood & discovery
python newmovies.py --mood "dark and intense"
python newmovies.py --mood "heist" --imdb-min 7.5
python newmovies.py --like "Parasite"
python newmovies.py --like "Inception" --mood "mind-bending"

# Sagas & franchises
python newmovies.py --saga "Star Wars"
python newmovies.py --saga            # auto-detect all incomplete sagas

# Filmographies
python newmovies.py --director "Stanley Kubrick"
python newmovies.py --actor "Al Pacino"
python newmovies.py --actor "Ben Stiller, Owen Wilson"     # any of them
python newmovies.py --cast "Ben Stiller, Owen Wilson"      # together only
python newmovies.py --cast "Robert De Niro, Al Pacino"
python newmovies.py --composer "Hans Zimmer"
python newmovies.py --author "Stephen King"
python newmovies.py --actor "Al Pacino" --artist-top 20   # top 20 only

# Collection intelligence
python newmovies.py --stats
python newmovies.py --analyze
python newmovies.py --analyze --no-timeout    # with large model
python newmovies.py --watchlist watchlist.csv

# Output options
python newmovies.py --actor "Pacino" --synopsis            # show plot before adding
python newmovies.py --actor "Pacino" --export pacino.html  # export to HTML
python newmovies.py --analyze --export gaps.csv            # export to CSV

# Tuning
python newmovies.py --imdb-min 7.5
python newmovies.py --sources 15 --suggestions 20 --top 15
python newmovies.py --sd 1960 --fd 1990                    # era filter

# Reset
python newmovies.py --resetblacklist

---

All options

Argument	Default	Description
--auto	off	Add all recommendations without prompting
--mood	off	Describe the atmosphere in plain language
--like	off	Base recommendations on any film title
--genre	off	Filter by genre (Sci-Fi, Horror, Comedy...)
--saga	off	Complete a franchise — specify name or use alone for auto-detection
--director	off	Missing films by a director
--actor	off	Missing films by actor(s) — comma-separated for multiple
--cast	off	Missing films where ALL listed actors appear together
--composer	off	Missing films scored by a composer
--author	off	Missing film adaptations of an author
--artist-top	0	Limit filmography results (0 = all)
--analyze	off	AI analysis of your collection + gap-filling recommendations
--stats	off	Collection statistics: genres, decades, ratings
--watchlist	off	Import from Letterboxd or IMDb CSV export
--synopsis	off	Show full plot synopsis when reviewing films one by one
--imdb-min	off	Minimum IMDb rating override (e.g. --imdb-min 7.5)
--export	off	Export to CSV or HTML (e.g. --export reco.html)
--no-timeout	off	Disable timeouts for large models
--sources	10	Source films sampled from your library
--suggestions	14	Ollama suggestions per source
--top	10	Final recommendations to keep
--score	6.5	Minimum IMDb rating (classic mode)
--score-relax	5.9	IMDb threshold in relaxed fallback
--sd	1970	Minimum release year
--fd	2030	Maximum release year
--no-embed	off	Disable plot embeddings (faster)
--resetblacklist	off	Clear the blacklist
--debug	off	Verbose output

---

Recommended models

Model	Size	Best for
llama3.1:8b	4.9 GB	Daily use, fast, good quality
mistral:7b	4.4 GB	Best balance speed/quality ⭐
mistral-small:22b	12 GB	Filmographies, best accuracy 🏆
llama3.2:3b	2.0 GB	Very fast, limited RAM

ollama pull mistral:7b
# then in config.yaml:
# ollama_model: mistral:7b

Use --no-timeout with 22b+ models:

python newmovies.py --analyze --no-timeout

---

Update

git pull

config.yaml, blacklist.json and logs are never overwritten.

---

If this is useful, a ⭐ on GitHub is always appreciated!