Prod.

Rated R Robot

Rollratedr 0.1.0
SceneNight Shift
Take1
Advisory

13+13+ for some language, violence and frightening sequences

DirectorYou
CameraYour browser
Date13 Oct 2026

A content advisory for screenplays

An AI-made film just got an R. Rated R Robot reads a screenplay the way a careful reader does: line by line, scene by scene.

It finds language, violence, frightening scenes, drinking and drugs, and sexual content. It gives each scene and the whole script an age band with the reason in plain words. Then it names the few lines that hold the script at that band, and shows you what happens if you cut them.

Running in this page: ratedr 0.1.0, an open-source library. Rule-based and transparent. Nothing you type leaves your browser. It is an advisory, not anyone's official rating.

Clap the slate to rate the next sample script.

The scenes of Night Shift, as rated

    Event 001

    Script desk

    Write or paste a screenplay in Fountain, the plain-text format screenwriters use. The page on the right is formatted and marked as you type.

    A scene heading starts with INT. or EXT.; a name in capitals on its own line is a character, and the lines under it are dialogue. A line in brackets under a name is a parenthetical.

    13+

    The band for each category
    CategoryBandWhy

    • LLanguage
    • VViolence
    • FFrightening
    • DDrinking, smoking, drugs
    • SSex and nudity
    Event 002

    Box office: $RATEDR

    The coin on this marquee. Live figures, the chart on the big screen, and your ticket.

    Now showing

    $RATEDR

    Price
    —
    Market cap
    —
    24 h volume
    —
    Liquidity
    —
    24 h change
    —

    The marquee lights up when the figures arrive.

    House lights down. The feature starts when the chart arrives.

    Feature presentation: 1-second candlesOpen the chart

    Admit one

    Contract address

    FZu9nxS72VUGqCzdezNbPNfbAduMpnoPfY3dUg8pump
    Supply
    1,000,000,000
    Tax
    0%
    Site
    v1.0
    Event 003

    The reel

    The whole script on one reel. Every spoke is a line, in order; every ring is a category.

    The reel

    Language is the outer ring and sex and nudity the hub. A gap in the reel is a cut between scenes. A filled cell is a finding: the deeper the colour, the stronger the finding. Point at a cell, or use the arrow keys on the reel, to read the line.

    Line

    Point at the reel.

      Event 004

      Cutting room

      Every marked line of the script hangs in the trim bin. Cut a line and the band is worked out again without it.

      Band now13+
      What would bring it down

          A trim marked holds the band is one of the fewest lines whose removal drops the script a band. Put back any one of them and the band returns.

          Event 005

          TV edit

          A broadcast version of the same script: the language is bleeped and everything else stays. Drag the wipe to compare the two cuts.

          Theatrical cutBroadcast cut 
          
      
      
    
          
  
          
  
  
          
    
    
    
  
          

          



          
  
          Event 006
          Reader's coverage
          The paper behind the library, filed the way a studio reader files a script.
          
  
          
    
          
      
          TitleReading for the Rating: A Transparent Content Advisory for Screenplays
          
      
          AuthorThe ratedr contributorsDateOctober 2026
          
      
          FormPaper, with a libraryLibraryratedr 0.1.0, MIT
          
      
          LoglineA rule-based reader gives a screenplay an age band, says why, and names the fewest lines that hold it there; every figure below is computed by the library on this page.
          
    
          
    
          
      Reader's grid
      
          ExcellentGoodFairPoor
          
      
        
          TransparencyX
          
        
          MethodX
          
        
          EvidenceX
          
        
          Limits statedX
          
                
    
          
    
          Consideras a first read, not a verdict
          

    
          Synopsis
          
    
          ratedr reads a screenplay written in Fountain and scans every line of action and dialogue for five kinds of content: language, violence, frightening scenes, drinking and drugs, and sex and nudity. Each category gives a band on a five-step scale, E, 10+, 13+, 17+ and 18+, and the highest band wins. Each scene gets its own band, and the reason is written in plain words. The library also names the drivers: the fewest lines whose removal drops the script below its band. On four sample scripts written for it, one per band from E to 17+, it gives the band each was written for. On 40 held-out lines written before the rules, it gets 39 exactly right.
          

    
          Comments
          
    
          1. Background
          
    
          In October 2026, several outlets reported that Gods Don't Give Gifts, an animated anthology of four stories made with generative AI by the YouTube creator Zack London and the AI studio Fable, had been rated R by the Motion Picture Association, and called it the first fully AI-generated film to get an MPA rating [2][3]. The reports also said its release had moved from 30 October to 4 December 2026, for a limited US run on fewer than 1,000 screens, so that it can be submitted for Best Animated Feature. One report adds that the Academy barred AI actors and AI-assisted scripts in May, while its rules say that AI tools "neither help nor harm" a film [2].
          
    
          Not connected. ratedr, this site and $RATEDR have no connection with that film, its makers, Fable, the MPA or the Academy, and none of them has seen this paper. ratedr has never read that film's script. Its bands are its own advisory and not anyone's official rating, and they are named differently on purpose.
          
    
          A rating is, in the end, a reading of what is on the page and on the screen. A writer usually wants to know early which band a draft is heading for and, more usefully, which lines decide it. That question has an exact answer for a set of written rules, and ratedr is built to give it.
          

    
          2. Reading the script
          
    
          Fountain is plain text with a few conventions [1]: a title page of Key: value lines, scene headings that start with INT. or EXT. (or a forced .), character names in capitals on their own line, dialogue under them, parentheticals in brackets and transitions that end in TO:. Notes in double brackets and the boneyard between /* and */ are dropped. The parser turns the script into scenes and typed lines; only action, dialogue and parentheticals are scanned, so a character called KILLER does not count as violence.
          

    
          3. Five categories, five bands
          
    
          
      Table 1. What moves a category up a band.
      
          Category10+13+17+18+
          
      
        
          Languagemild wordsone strong word, or any moderatetwo or more strong words
          
        
          Violencea shove, a punchweapons, shooting, blood, killinggraphic: severed, entrails, spraying blood
          
        
          Frighteningone or two scary momentsthree or more
          
        
          Drinking, drugsdrinking or smokingdrug referencesdrug use
          
        
          Sex and nuditysuggestive contentnudityexplicit content
          
                
    
          
    
          A kiss is E. Words that only look violent or rude are left alone: a CLOSE SHOT, "kill the engine", "dead tired", a blood orange, "shoot the photos", smoke from a chimney. The word lists are short, and they sit at the top of the library, so changing a list changes the advisory with it.
          

    
          4. Scenes and the reason
          
    
          The same judgement runs on every scene, so a 13+ script can show which scene carries it. The reason lists the categories that matter, worst first, in words a parent would use: "13+ for some language, violence and frightening sequences". When a script is 13+ or higher, categories at 10+ are left out of the reason, as they no longer decide anything.
          

    
          5. What holds the band
          
    
          For each category at the top band, ratedr works out how many findings at the deciding level must go (all of them, or all but one strong word, or all but two scary moments) and takes the lines that carry the most of them. The result is the drivers. The tests check, for every sample, that cutting the drivers drops the band and that putting back any single one of them restores it. Repeating the step gives a path down: Salvage goes from 17+ to 13+ and then to E.
          

    
          6. The broadcast edit
          
    
          clean() masks strong and moderate language (mild words too, if asked) and leaves the rest of the script as it was. After the default edit no strong or moderate word is left, so the language of Salvage drops from 17+ to E and that of Night Shift from 13+ to 10+, where its mild words remain; with mild words masked too, the language of every sample is E. The violence, the fear and the drinking stay, because bleeping does not remove them.
          

    
          7. Evaluation
          
    
          Three sets, kept apart. Four sample scripts were written for the library, one per band. 53 development lines are the lines the rules were written against; they are all right (53 of 53), which says only that the rules fit them. 40 held-out lines were written before the rules and never used to tune them.
          
    
          Table 2. The sample scripts, rated on this page.
          ScriptWritten forRatedReason
          
    
          Table 3. The held-out lines, scored on this page.
          CategoryLinesPrecisionRecallLevel right
          
    
          The one miss is a line that says a character stabs a creature: the word creature also counts as a frightening moment, which the person who labelled it did not intend.
          

    
          8. Where it fails
          
    
          The same people wrote the rules, the lines and the scripts, and the lines are short and clear, so real scripts will do worse. Sarcasm, slang, context ("he kills it on stage"), violence told through dialogue that the word lists do not know, a scene that is frightening without a frightening word, and anything not in English will be missed or misread. Counting is crude: two strong words in a script of a hundred pages weigh the same as in a script of two. Nothing here looks at images, sound or performance, which is where much of a real rating is decided.
          

    
          9. Use
          
    
          Use the band as a first read for a human: to see early where a draft is heading, to find the lines a note is really about, or to make a broadcast version of the dialogue. Do not use it in place of a rating board, a broadcaster's standards team or a parent's own judgement.
          

    
          References
          
    
            
      
          1. Fountain, the plain-text screenplay format. fountain.io
            2. 
      
          2. F. Landymore, "First fully AI movie gets MPA rating, eyes Academy Award", Futurism, 8 October 2026. futurism.com
            3. 
      
          3. D. Stojkov, "YouTuber Gossip Goblin moves AI film Gods Don't Give Gifts to December for Oscar campaign", NetInfluencer, 7 October 2026. netinfluencer.com
            4. 
    
          
    
          Coverage by the ratedr contributors. Recommendation: consider, as a first read.
          
  
          

          



          
  
          Event 007
          The vault
          ratedr 0.1.0: one file, no dependencies, Node 18 and later and every modern browser. MIT licence. Every can is a file of the repository, with its SHA-256.
          
  
          
    
          
      
          Downloads
          
      
          
      
          Quick start
          

          const R = require("./ratedr.js");
// in a browser: <script src="ratedr.js">

const text = fs.readFileSync("script.fountain", "utf8");
const r = R.rate(text);
r.band;           // "13+"
r.reason;         // "13+ for some language, ..."
r.drivers.lines;  // the lines that hold the band

R.rate(text, { trim: r.drivers.lines }).band;  // "10+"
R.clean(text).text;  // language bleeped
          

          node bin/ratedr.js rate script.fountain
node bin/ratedr.js whatif --sample salvage
node bin/ratedr.js clean script.fountain --mask "***"
node bin/ratedr.js eval --misses
npm test
          
    
          
    
          
      
          The film cans
          
      
            
    
            
  
            
  
  
            
    
            Credits: the API