Open SourcePython / LangGraph / Textual

Film Pipeline: structured state for AI-assisted film creation

Film CreationLangGraphQC
View Repository
ProblemCreative agent workflows lose state, approvals, dependencies, and validation evidence.
ShapeLangGraph pipeline with human gates, typed state, artifacts, and a TUI.
StatusIn active development with CI and product gates.

The Film Pipeline takes an idea through structured pre-production, generation planning, QC evidence, and validated handoff state. It is not trying to be final editing software. It is the orchestration and evidence layer before downstream generation and finishing.

The project treats creative work as stateful production, not a sequence of disconnected prompts. Artifacts are versioned. Approvals are explicit. Repairs preserve dependency history. Validators produce evidence that can be inspected.

Why I built it

AI film workflows often fail because the process is invisible. Chat history becomes the production system. That does not scale when a scene, matrix, guide, asset plan, or QC issue needs to survive revisions and handoffs.

What to inspect

  • Typed LangGraph state contract with append-only reducers.
  • Human approval gates using real graph interrupts.
  • Versioned artifacts with dependency tracking.
  • Parallel QC validators using LangGraph fan-out.
  • Terminal operator console for project control.

Why it belongs here

This project shows the same architecture discipline in a creative domain: state contracts, checkpoints, validation, rollback, and inspectability. The medium is film creation, but the engineering problem is reliable agent workflow design.