Screenplay development system

A screenplay development system for a feature-film adaptation, built under contract for Sony through a professional screenwriter. It gave the writing team one organized, versioned home for the script and kept AI writing assistants consistent across a 110-plus page draft.

This work is under NDA, so I can't name the project or show its materials. What follows describes the system I built, not the story.

Role
Contract developer
Timeline
October to December 2025
Client
Sony, via a professional screenwriter
Stack
Git, Fountain, JSON, Markdown, Python, HTML/CSS/JavaScript, bash
Status
Delivered, reused as a template

The problem

Adapting existing source material into a screenplay means tracking a huge amount of detail: characters, timelines, what's been set up and what still needs to pay off. Writers were also using AI assistants to help draft and review, and those assistants forget everything between sessions. Without a system, continuity slips and the same questions get answered over and over.

What I built

  • A Git-versioned development system with a project bible, a database of the source material, Fountain templates for acts and scenes, and more than 10 continuity and quality-check checklists. The screenwriter and collaborators used it daily.
  • A "context engine" in JSON and Markdown that tracks rolling context, scene dependencies, setups and payoffs, and each character's state. It kept AI writing assistants consistent across a script of more than 110 pages.
  • A 71-page knowledge-hub website in HTML, CSS and JavaScript, so the whole team could browse the project's reference material.
  • A presentation package for development meetings: a beat sheet, a character compendium and a roadmap.
  • A Python exporter that turns Markdown documents into clean PDFs.

Hard parts

Git for people who aren't programmers

Version control is the right tool for a script that many people touch, but writers shouldn't have to learn developer workflows. I set up role-based branching, wrote bash setup scripts that do the configuration for them, and made HTML onboarding guides that walk through everyday tasks in plain language.

Making AI assistants remember

An AI assistant only knows what you give it. The context engine gives it the right slice of the project each time: what came before this scene, what it depends on, and where each character stands. That's what kept its suggestions consistent from the first page to the last.

Results

  • Used daily by the screenwriter and collaborators throughout the contract.
  • AI-assisted drafting stayed consistent across a 110-plus page script.
  • Packaged as a reusable template, version 2.2, and redeployed for a second novel-to-screenplay project.

Where it stands today: the contract ended in December 2025, and the system lives on as a reusable template, version 2.2, already used on a second project.

Need a system for a complex creative or knowledge project?

I build tools that keep teams and AI assistants organized. Let's talk about yours.

Book a free call

See the custom tools service Next: AI and automation Back to all work