FlywheelApp · full
Cinema + labs + WebGL. The labeled map, Hub canvas, field sketch, and every live engine. Use when this is the app that owns the GPU on the machine.Every feature, no graphics. Same engines, same API, same honesty grammar. No WebGL, no custom cursor, no Three.js. For when other apps already hold the GPU, or after learning loops when you only need the work.Terminal + one API out. Every live verb as a command and as HTTP. TUCE agents load skill.md and call /api/v1. No cinema overhead. Occupancy stays nominated | live-bound | unavailable.

Setup · Goals

Say the outcome before you pick a model.

Goal-first setup is how the hub avoids catalogue-driven work. You name what should happen. The hub queues tasks. You still have to start a cycle.

Goal-first

A saved goal is not a running spoke.

The picker can seed a builder backlog. Start MoA Cycle is still the live verb.

How it connects

Follow the occupancy.

documented

Name the outcome. Queue the work. Then start a cycle.

Goal-first setup is not a model picker in disguise. The hub asks what you want before it names a spoke.

  1. 01Goalsoutcome first
  2. 02Tasks queuedbacklog, not a run
  3. 03Run composeroccupancy attached
  4. 04Start MoA Cycledemo stays labeled

Demo MoA is a sandbox and stays labeled. Live MoA for a signed-in operator is a different path. A FAST seed cannot silently become RESEARCH.

Why it exists

Fit follows purpose.

codingIntent classifier labels coding before a model is picked. Adapter plates such as code-r16 attach on L08. Fit still gates the base. A label is not a silent model override.documentsDocument / retrieval-shaped outcome. Window cacher (L04) and episodic recall (L05) are the live tools on this host. A saved goal queues tasks — Start is still the live verb.researchRESEARCH is also an L0 mode, and that is a different word. This kind names a wider complementary portfolio. FAST cannot silently become RESEARCH, and it cannot commit.course creationCurriculum-shaped outcome. Saving the goal can seed the builder backlog. Occupancy still has to be earned on a cycle.visionNominates L11. Field sketch reads a local safetensors header and an optional 2 MiB sample. Not PCA. Not a claimed VLM bind.image generationGeneration-shaped outcome. Occupancy stays unavailable until a spoke is live-bound. This host will not invent a sampler.voiceNominates L14 acoustic. No speech spoke is live-bound on this marketing host. Stay unavailable rather than fake occupancy.data scienceTabular / analysis outcome. Constraint slot (L10) and the context pack are the live tools. Tokens/sec are not published.formal mathSAT path. Clause mill, watched DPLL, JSON constraints. Cap 24 vars / 80 clauses. Capped ≠ UNSAT. SAT ⟂ HAL.automationTool-calling outcome. Scan lists fingerprints; Accept is still required per tool. Deny is the default. Scan ≠ grant.private chatDialogue slot (L06). Token completion on this host still refuses. Point TUCE at a live Hub when a model spoke is live-bound.custom integrationCaller / CLI / skill.md pack. One API out. TUCE remains TUCE. Occupancy travels with the answer or the refusal.

Purpose

What should happen

A goal is an outcome label. It is not a secret model override.

Queue

Tasks get seeded

Saving a goal can queue local tasks or open Workspace so you can sync the builder backlog.

Start

You press Start

Nothing live-binds because a goal was saved. Occupancy still has to be earned.

A saved goal is not a running spoke. Hover a kind for the occupancy slot and live tool it maps to on this host.

← SolversSAT console and StrataX. A solve is not a hardware bind.All integration flowsEvery Hub menu as a flow: scan, fit, bind, hot-swap, StrataX, observe, upgrade.Downloads →Pull is not bind. Fit still gates activation.