docs: surface v1.8 substrate capabilities in README + CHANGELOG

- README: dual-band bullet updated (now real at the Value level, not just JIT); new
  "Substrate-into-core (v1.8): identity is content" capability bullet; SUBSTRATE.md in the
  documentation map.
- CHANGELOG: v1.8.0…v1.8.3 timeline entry.

No code changes; version bump to 1.8.4. 172/172 core tests pass.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: The Architect

CHANGELOG.md

@@ -13,7 +13,7 @@ Read top-to-bottom for the arc; jump to any chapter for the detail.
 
 | Tag | Date | One-line |
 |---|---|---|
-| [v0.8.4-substrate-builtins](#v084-substrate-builtins--2026-05-17) | 2026-05-17 | **40× CPU / 96× GPU end-to-end speedup on Prometheus**. Fused `substrate_adamw_update` Rust builtin replaces ~15 OMC-side element-wise loops per parameter — what was 25.81 s/step at d_model=256 is now 0.65 s (CPU) / 0.27 s (GPU). The v0.8.2 GPU integration and v0.8.3 substrate-tile win finally pay out end-to-end. Identical training trajectory. |
+| v1.8.0 … v1.8.3 | 2026-05-30 | **Substrate promoted into the core language.** New modules `address`/`cas`/`locality`/`synth`; ~22 builtins + `@memo`/`@dualband` pragmas. Content-addressing (`haddr`, χ²≈9), addressable heap + persistent `@memo` (`fib(90)` instant, cross-run), O(1) semantic equality (`same_value`), locality similarity (recall 0.99 vs φ 0.02), verify-gated self-modification (`fn_swap_verified`), correct-by-construction synthesis (`gen_omc`, parse/run 1.000 over 300), HBit dual-band **real at the Value level** (`phi_shadow`/`bands`/`value_divergence`). NEXT-7 answered on CPU: capability scales with addressed content at flat per-query cost. 172/172 core tests pass. See [`SUBSTRATE.md`](SUBSTRATE.md). |
 | [v0.8.3-substrate-gpu](#v083-substrate-gpu--2026-05-17) | 2026-05-17 | **Substrate-shaped GPU matmul wins +38% vs conventional 16×16**. Anisotropic 8×32 tile (Fib short dim, wavefront-divisor long dim) hits 114 GFLOPS at 1024² vs 71 for the standard tile. Pure-square Fib tiles (13×13, 21×21) still lose; the win comes from substrate suggesting "8 first" + hardware demanding wavefront alignment. New default tile baked into the CLI integration. |
 | [v0.8.2-gpu-prometheus](#v082-gpu-prometheus--2026-05-17) | 2026-05-17 | **GPU wired into Prometheus** via a MatmulAccelerator hook. **13× speedup on synthetic chained matmul** (512², CPU 3.47s → GPU 0.27s). End-to-end Prometheus training at d_model=256: wall-clock unchanged — OMC tree-walk overhead in substrate-shaping helpers (smod, resample, Q6) is the next bottleneck, not matmul. Integration is load-bearing for the substrate-native GPU kernels coming next. |
 | [v0.8.1-tape-primitives](#v081-tape-primitives--2026-05-17) | 2026-05-17 | **Substrate-native tape primitive precedent**: `tape_phi_log` fuses Q6's log-distance into one tape node, with `tape_abs` as the boring companion. Composed vs fused trains to within ~1e-7 — fused abstraction is free. **Pre-existing tape_div/tape_mul broadcast-backward bug fixed**, which unblocks OMC-side cross-validation of S-MOD + substrate-K. First Q6 OMC replication: −0.63% 2/3 seeds at small scale, directionally matching PyTorch's −12.15%. |

Cargo.lock

@@ -21,7 +21,7 @@ exclude = ["omnimcode-python"]
 resolver = "2"
 
 [workspace.package]
-version = "1.8.3"
+version = "1.8.4"
 edition = "2021"
 authors = ["The Architect <architect@sovereign-lattice.io>"]
 license = "MIT"

Cargo.toml

@@ -34,7 +34,9 @@ These are concrete language features, not aspirations:
 
 - **The substrate is a primitive, not a library.** `HInt`, OMC's integer type, carries a φ-resonance and HIM score computed at construction. Every `Value::HInt(_)` ever created has been routed through `compute_resonance` and `nearest_attractor_with_dist`. Substrate-ness is at the type level.
 
-- **Dual-band executable code.** OMC values have a classical α-band and a harmonic shadow β-band, packed into LLVM `<2 x i64>` SSE2 vectors inside JIT'd functions. `phi_shadow(x)` makes β diverge; `harmony(x)` reads the substrate-routed coherence. **Branch elision based on harmony** is shipped: high-coherence inputs skip entire conditional blocks at native code speed (95.2% reduction on high-harmony inputs).
+- **Dual-band execution — now real at the Value level (v1.8).** OMC values carry a classical α-band and a harmonic shadow β-band. `phi_shadow(x)` attaches β (= nearest Fibonacci attractor); it then **rides through arithmetic** (`bands(x)` → `[α, β]`), and `harmony(x)` / `value_divergence(x)` read the accumulated drift — α is always the exact answer. In JIT'd functions the bands are packed into LLVM `<2 x i64>` SSE2 vectors, and **branch elision based on harmony** is shipped (95.2% reduction on high-harmony inputs).
+
+- **Substrate-into-core (v1.8): identity is content.** A value's address is a function of what it *is*. `@memo` gives transparent, **persistent, cross-run** memoization (`fib(90)` instant); `cas_put`/`cas_get` + `same_value` are a content-addressed heap with O(1) semantic equality; `locality_fp`/`nearest_fn` add content-similarity retrieval; `fn_swap_verified` is self-modification the interpreter must verify before it accepts; `gen_omc` emits valid-by-construction OMC (parse/run 1.000 over 300 seeds). Capability scales by **adding addressed content at flat per-query CPU cost** — see [`SUBSTRATE.md`](SUBSTRATE.md).
 
 - **O(log_φπfib N) algorithm family.** `substrate_search` and friends use F(k)/φ^(π·k) split-points — each iteration shrinks the live range by **φ^π ≈ 4.534**, not 2. The canonical iteration bound is `log_φπfib(n) ≈ 0.459 · log₂ n`. A complete primitive family is exposed: `substrate_lower_bound`, `substrate_upper_bound`, `substrate_rank`, `substrate_count_range`, `substrate_slice_range`, `substrate_intersect`, `substrate_difference`, `substrate_insert`, `substrate_quantile`, `substrate_select_k`, `substrate_nearest`, `substrate_min_distance`, `substrate_hash`.
 
@@ -246,6 +248,7 @@ The MCP server + substrate codec are the entry points designed for LLM agents dr
 | [`omnimcode-core/src/prometheus/README.md`](omnimcode-core/src/prometheus/README.md) | Substrate-native ML framework |
 | [`experiments/prometheus_parity/`](experiments/prometheus_parity/) | Substrate-attention findings (K, S-MOD, V), each with a `FINDING.md` |
 | [`docs/SUBSTRATE_NATIVE_AGENT.md`](docs/SUBSTRATE_NATIVE_AGENT.md) | Two-agent demo composing every substrate primitive |
+| [`SUBSTRATE.md`](SUBSTRATE.md) | **Substrate-into-core primitives (v1.8.x):** content-addressing, addressable heap + `@memo`, locality similarity, verify-gated self-modification, `gen_omc` synthesis, value-level dual-band — verified numbers + honest limits |
 | [`CHANGELOG.md`](CHANGELOG.md) | Chapter-by-chapter project history (mirrors the release notes) |
 | [`ROADMAP.md`](ROADMAP.md) | What's planned next |