Claude Code Doc Review - 2026-05-27
The user's correction was right. Several generated docs treated intended design language as if it were verified runtime behavior. The source supports a strong architecture, but the language must stay precise:
Full Public Reader
Claude Code Doc Review - 2026-05-27
Verdict
The user's correction was right. Several generated docs treated intended design
language as if it were verified runtime behavior. The source supports a strong
architecture, but the language must stay precise:
- "Diffusion" is a legacy service name for a conditioned one-step flow/token path.
- SAN is wired and weight-backed locally, but not proven as the sole audio driver.
- N'Ko ASR is anchored by trajectory-biased Transformer CTC, not MAOE routing.
- MAOE is a post-ASR correction/admissibility/control layer.
- The 128D vector is a family of related runtime buffers with known shims and
mismatches, not a perfectly uniform contract everywhere yet.
Source-Grounded Corrections
| Topic | Incorrect generated claim | Verified source evidence | Correct wording |
|---|---|---|---|
| Diffusion | The app uses a full multi-step diffusion sampler over the full 128D body state. | `MotionMixApp/Services/DiffusionService.swift` says the name is historical and the current path is ConditioningEncoder + FlowGenerator1Step. CoreML signatures are `dynamics [1,104] -> embedding [1,768]` and `noise [1,384,81] + conditioning [1,768] -> logits [1,384,81]`. | Conditioned one-step flow/token generation path with hub and rule-based fallbacks. |
| 104D/128D | There is one universal 128D vector consumed identically everywhere. | `DiffusionService` builds 128D but truncates to 104D for `ConditioningEncoder`; Rust SAN defaults to 128D; some comments still say 104D; ClaimBridge comments say 104D while FFI reads 128 floats. | Treat dynamics as related source-specific vectors with explicit producer/consumer boundaries. |
| SAN training | SAN was trained to understand BodyTruth modes and all modality masks. | `SANService` loads `san_weights.bin`/`san_manifest.json`, calls `san_step`, and logs flat inputs. No source inspected proves BodyTruth-mode training. | SAN is a Rust/Swift adaptive output pipeline with local weights and trajectory logging; training claims require named artifacts. |
| SAN audible control | SAN drives music after calibration. | `SANService.mixFactor` defaults to `0.0`; Rust `SANConfig::default()` uses `mix_factor = 0.0`. | SAN affects audio only when weights load, the consumer blends SAN output, and mixFactor is above zero. |
| FAN/NHA/TTT | These layers have proven learned selectivity, rhythm learning, and fixed-time calibration. | Source implements/configures FAN, MoE, NHA, TTT machinery, but behavior must be proven by weights and runtime logs. | Describe the implemented modules and require artifacts/logs before claiming learned performance. |
| N'Ko ASR | MAOE is the trained ASR model or acoustic anchor. | `nko-brain-scanner/HANDOFF.md` identifies N'Ko Trajectory CTC, UnifiedCTCHead, Whisper large-v3 encoder features, 6-layer Transformer with trajectory bias injection, 20.57 | |
| LIM-RPS | LIM-RPS is the canonical deployed runtime name. | Source names include `EchelonCore`, `LatentUpdater`, `SANPipeline`, `DiffusionService`, and `ClaimBridge`; LIM-RPS appears as historical/research vocabulary and in specific files. | Use LIM-RPS as constraint/aesthetic vocabulary unless citing a concrete implemented path. |
Files Corrected In This Pass
- `LUME-CC/03-body-sensing/body-truth.md`
- `LUME-CC/01-foundation/the-recursive-loop.md`
- `LUME-CC/02-the-stack/data-flows.md`
- `LUME-CC/09-reference/glossary.md`
- `LUME-CC/00-INDEX.md`
- `computational-choreography/04-generative-output/audio-engine.md`
- `computational-choreography/00-INDEX.md`
- `computational-choreography/09-reference/claim-audit-2026-05-27.md`
Language Rules Going Forward
Use these as hard constraints when updating the docs:
- Do not say "multi-step diffusion" for MotionMixApp unless a new sampler is
actually wired and cited.
- Do not say SAN "learned" or "took over" unless a training artifact and runtime
blend path are cited.
- Do not say BodyTruth feeds every output path unless the current integration is
named.
- Do not call MAOE the ASR anchor. The verified N'Ko anchor is trajectory-biased
Transformer CTC.
- Do not collapse movement inscriptions, ASR correction, AirDeck gestures, and
SAN music regimes into one term. They are related layers, not the same object.
Current Accurate Architecture Sentence
LUME's current source-grounded spine is: body and sensor streams feed Echelon /
BodyTruth-style state; SAN, DiffusionService, Unity visuals, AirDeck, and
inscription layers consume bounded state through separate integration paths; N'Ko
provides both a verified ASR research lane and a motion-inscription memory lane;
MAOE-style routing belongs around correction/admissibility, not as the acoustic
ASR core.
Promotion Decision
Attach run IDs, datasets, metrics, and reproduction commands.
Source Anchor
computational-choreography/09-reference/claude-code-doc-review-2026-05-27.md
Detected Structure
Method · Evaluation · Figures · Code Anchors · Architecture