# ExergyNet External Claims and Trade-Secret Boundary Register
# EXERGYNET_EXTERNAL_CLAIMS_AND_TS_BOUNDARY_REGISTER.md
# Status: ACTIVE -- Updated 2026-09-02
# Rule: PUBLISH THE PHENOMENON. PROTECT THE RECIPE.

## Boundary Classification

### PUBLIC (unrestricted disclosure)
- Measured result (with stated envelope and limitation)
- Validated experiment scope
- Architecture concept (layer/function, not mechanism)
- Component name and responsibility
- Public principles (MEMORY GROWTH != INFERENCE GROWTH, RIGHT TO KNOW != RIGHT TO ACT)
- Benchmark summary numbers (see claim ledger for individual claim status)
- Settlement subsystem documentation (LNES-03, LNES-04 status, contract addresses)

### NDA (available under non-disclosure)
- Deeper methodology for LNES-82C and H200 campaigns
- Full statistical distributions beyond public summary
- Workload specification details beyond summary
- Detailed architecture tradeoff analysis
- LNES-86 integration approach (high level, not mechanism)

### EDT_TS (trade secret -- not to be disclosed)
- Work Compression rarity thresholds
- Work Compression prefix construction rules
- Posting logic and pruning rules
- Crossover thresholds for routing decisions
- Cache topology internals
- LNES-86 adaptive retrieval mechanism (selectivity emergence mechanism)
- TS security operational mechanics
- Settlement private key management
- Internal scoring coefficients

## Current External Claim Status

| CLAIM | CURRENT_STATUS | ACTION |
|-------|----------------|--------|
| ExergyNet is a decentralized protocol | STALE/INCORRECT | FAQ, homepage, llms.txt rebased |
| dual-layer settlement engine | HISTORICAL | Rewritten in FAQ as historical context |
| Layer-0 Mesh / Layer-1 Membranes | SUPERSEDED | Labeled historical in FAQ |
| Proof-of-Exergy / PoX | HISTORICAL | Labeled historical in FAQ |
| $EXG as current accounting token | LEGACY | token.html marked Historical; FAQ updated |
| Solana/Base as primary category | STALE | Repositioned as settlement subsystems |
| RISC Zero as universal compute | NEEDS_CONTEXT | Existing note present; no overgeneralization added |
| 8K->285K as primary benchmark | CURRENT (H200) | Correctly labeled as H200 campaign |
| 660-820 tokens | CURRENT (H200) | Correctly labeled as H200 campaign |
| 125x corpus growth to 4M | CURRENT | Primary scaling evidence; claim ledger CLM-005 |
| ~903 mean K at 4M holdout | CURRENT | Claim ledger CLM-007 |
| 11.3x correct-task throughput | CURRENT | Claim ledger CLM-001; envelope stated |
| USENIX Security 2026 | EXTERNAL_CITATION | Softened to citation; external paper (Wang et al., arXiv:2607.19545) |
| AutoHunter / Agent #116 / 0xWork | UNVERIFIED_CURRENT | Retained in docs; not presented as primary architecture |
| MyMonitor.ai | UNVERIFIED_CURRENT | Appears as logo; retain pending corporate status review |

## Disclosure Principles

1. Publish the phenomenon (what was measured, what was observed)
2. Protect the recipe (how it works mechanically, thresholds, construction rules)
3. State every envelope (hardware, corpus, workload scope)
4. Disclose every material limitation in the same breath as the result
5. Separate campaigns (H200 and A100 are independent; do not merge absolute K values)
6. Label historical content explicitly (HISTORICAL ARCHITECTURE / SUPERSEDED DESIGN)

## Revision History

- 2026-09-02: Initial register created; LNES-82C semantic rebase documented