CMG learning field guide / research build
Learn the system
around the model.
Each lesson crosses the same boundary: concept → worked shape → deterministic demonstration → executed evidence → understanding check.
Concept→Shape→Run→Evidence→Check
00–08 / curriculum
00
Start with the system shapeA model is one component. The graph exposes the system around it.
Open lesson →
01
Run one small primitive end to endThe primitive is deliberately small enough to understand by inspection.
Open lesson →
02
Treat a forward pass as ordinary transformsA dense layer can be inspected through the same primitive as any other state transition.
Open lesson →
03
Turn error into a directed signalError says what is wrong. Sensitivity says where change has leverage.
Open lesson →
04
Map a lifecycle without importing its domainObserve, measure, evaluate, and decide has the same system shape as a neural feedback loop.
Open lesson →
05
Ask whether quality earns its costUseful flow compares configurations; sensitivity points to the next knob.
Open lesson →
06
Decode a coherent path through noiseThe best state at every step is not always the best path through all steps.
Open lesson →
07
Project execution without inventing structureThe viewer derives from the graph and completed trace instead of maintaining a second diagram.
Open lesson →
08
Keep findings separate from decisionsA structural observation is evidence. The caller decides what it means.
Open lesson →
Nine executable lessons.
Supporting material
Field notes.
Field note / philosophy
Theory in the bones
The surface stays small enough for two different engineers to use after a short investigation.
Field note / development
How a capability earns its place
Real use case, language parity, general surface, inspectable result, and a reasoning trail.
Field note / structure
Two languages, one design
Language-agnostic design lives once; TypeScript and Python implement it in lockstep.