Barre Code: Movement Cipher Lab

Dance the algorithm. Decode the movement.

Comment

What it is

Barre Code: Movement Cipher Lab is a standalone embodied-computing lesson. Learners arrange six synthetic movement ciphers, compile their structural logic, find a seeded glitch, change one bounded parameter, and download a Learning Receipt.

The central authority rule is simple:

Code validates. AI explains. The human keeps moving.

Deterministic TypeScript decides structural correctness. GPT-5.6 Sol is an optional, server-only explanation layer that runs only after the sequence has already passed validation. Model failure never blocks lesson completion. No human movement is captured, scored, diagnosed, or evaluated.

Inspiration

Computational thinking is often introduced at a keyboard, even though sequence, state, parameters, repetition, debugging, and resolution already exist in movement instruction. Movement Cipher Lab makes those relationships legible without turning a body into a dataset.

The educational premise builds on Allison Van Cura's earlier Ballet Barre Code and Ballet Bots research. The complete public lesson implementation, compiler, API boundary, tests, responsive interface, and receipt were newly built during OpenAI Build Week.

How the lesson works

  1. Decode six movement ciphers and their computational ideas.
  2. Arrange them by dependency using drag-and-drop or accessible earlier/later buttons.
  3. Compile the phrase against five deterministic TypeScript rules.
  4. Glitch a deliberately incorrect sequence and identify the broken dependency.
  5. Coach with an optional GPT-5.6 Sol explanation of the already validated result.
  6. Create a bounded variation by changing one quality parameter while preserving structure.
  7. Prove completion with a downloadable Learning Receipt that clearly records its limits.

Technical implementation

The app uses the Next.js App Router with TypeScript and React. Pure modules own the movement schema, deterministic validator, scoring, glitch detection, bounded variation rules, and receipt construction.

The server-only /api/coach route revalidates the submitted sequence before any OpenAI call. Invalid structures return HTTP 422 and never reach the model. Valid results use the OpenAI Responses API with strict JSON Schema structured output. Missing credentials, quota errors, malformed output, and other model failures return deterministic fallback coaching with HTTP 200.

The browser interface includes accessible buttons in addition to drag-and-drop, large mobile touch targets, visible keyboard focus, semantic HTML, and reduced-motion support.

How Codex accelerated the build

Codex converted a precise repository constitution and founder-authored movement pedagogy into a tested product under a compressed Build Week window. It:

  • consolidated the lesson into one bounded golden path;
  • implemented the pure movement schema, compiler, and Learning Receipt modules;
  • enforced the deterministic-before-model authority boundary;
  • built the responsive interface and server-only Responses API route;
  • wrote boundary tests for valid sequences, glitches, variations, malformed model output, synthetic-only fixtures, and receipt limitations;
  • ran lint, TypeScript validation, nine automated tests, production builds, deployed runtime checks, and a client-bundle secret scan;
  • preserved clean provenance through three incremental Git commits and a dedicated feature branch.

Verification

  • ESLint: passed
  • TypeScript: passed
  • Tests: 9/9 passed
  • Next.js production build: passed
  • Vercel production build: ready
  • Deployed root: HTTP 200
  • Valid sequence with unavailable model: HTTP 200 deterministic fallback
  • Invalid sequence: HTTP 422 before model access
  • Client bundle secret scan: passed

Data and claim boundaries

The lesson uses synthetic data only. It captures no image, audio, video, biometric, health, identity, client, or student information. It makes no physical-safety, medical, therapeutic, validated artistic-scoring, accreditation, encryption, blockchain, token, wallet, NFT, investment, or production-readiness claim.

The Learning Receipt records lesson logic only. Its evidence identifier is a deterministic checksum, not secure encryption or a credential.

Built during OpenAI Build Week

  • the complete Movement Cipher Lab lesson flow;
  • deterministic movement compiler and scoring;
  • glitch and bounded-variation challenges;
  • server-only structured GPT-5.6 Sol coaching;
  • deterministic model-failure fallback;
  • Learning Receipt construction;
  • responsive accessible interface;
  • automated boundary and logic tests;
  • documentation, deployment, and submission materials.

Pre-existing research

The movement-as-computation teaching premise comes from Allison Van Cura's prior Ballet Barre Code and Ballet Bots research. Private research archives, client material, and the full MetaCodeography source are excluded from this public repository.

Built With

Share this project:

Updates