Quick definition
Vibe coding — a term from OpenAI researcher Andrej Karpathy — means describing what you want in plain language, letting an AI generate the code, and steering by feel and results instead of reading and understanding every line yourself. Scene Doctor, the tool now live at /tools/scene-doctor/, was built this way: one afternoon, zero programming background, working directly with Claude as a collaborator through each step from front end to deployment.
In one line: The distance between "I wish this existed" and "I built a rough version of this" has collapsed.
Why this counts as a skill writers already have
Vibe coding doesn't require learning a programming language — it requires describing a problem clearly, noticing when something feels wrong, and asking better questions until it doesn't. That's the exact same skill set exercised daily in a writers' room breaking story: nobody there is executing the craft alone either, and the room's value comes from clear description and iterative refinement, not solo technical mastery.
The build, step by step
| Step | What happened |
|---|---|
| 1. Strip to the simplest setup | A small HTML widget on the site, paired with a lightweight serverless backend — no heavy app, no ongoing maintenance. |
| 2. Build the front end | AI-generated HTML, CSS, and JavaScript — layout, email gate, character counter, loading states. |
| 3. Build the backend | The writer's own scene methodology (WANT, Scene and Sequel, escalation through resistance) translated into a system prompt. |
| 4. Deploy | Guided, step-by-step, through hosting setup and environment variables — copying and pasting commands with the reasoning explained at each stage. |
The tool's real ingredient wasn't code — it was the methodology
What made the backend genuinely useful wasn't generic AI notes; it was the writer's own accumulated frameworks — WANT, Scene and Sequel, escalation through resistance, reaction, and new tactics — translated directly into a system prompt. A post can explain a framework in the abstract. This tool applies the same framework to a specific scene at 3 a.m. when nothing else is available to say "this is where it's breaking, and this is why." The AI supplied the code; the writer supplied the diagnostic substance that made the code worth having.
What actually got fixed along the way
When the paste function failed specifically in Safari, the collaborating AI diagnosed and fixed the browser-specific issue in under a minute — a bug the writer had no way to independently understand, let alone resolve. This is the practical shape of vibe coding: not needing to understand every failure, but always knowing what each step is supposed to accomplish and how to recover when it doesn't.
What it actually cost
Hosting: free, on a serverless tier. API usage: pennies per run. Time: one focused afternoon. Programming knowledge required: almost none — editing a single config file in a plain text editor was the technical ceiling.
Vibe-coding checklist
- Can I describe the problem this tool should solve in one clear paragraph?
- Am I treating the AI as a collaborator I question, or a black box I blindly trust?
- Does the tool encode something only I know (a methodology, a voice), or is it generic?
- Have I picked the simplest possible setup, or am I building more than the problem needs?
Try the tool this essay describes
Scribbler’s Scene Doctor
Use the Scene Doctor — the actual tool this page is the origin story of — to analyze a scene using the WANT and Scene/Sequel frameworks described above.
Source note
This maintained reference page develops ideas first published in “Vibe Coding an AI Scene Analyzer” in Scribbler’s Toolbox on January 10, 2026. The original essay remains part of the editorial archive; this version is structured as a durable craft reference.
Related pages
- AI Script Doctors: Building a Team of Specialized AI Editors for Your Screenplay — extending the same collaborate-with-AI approach into a full room of specialized reviewers.
- The WANT Framework for Screenwriting: How to Give Every Character a Playable Scene Objective — the actual methodology that became Scene Doctor's system prompt.
About the author
Jesse Alexander is a screenwriter, showrunner, and narrative designer whose work spans film, television, games, and transmedia storytelling. Scribbler’s Toolbox collects practical frameworks developed from professional story development and production.