V121: deep debt + ecosystem absorption — smart refactoring, lint hardening, docs/handoff sync

Structural: biomeos/mod.rs 631→232 lines (extracted storage, compute, registration,
health, routing, client_tests submodules), stats/agreement.rs → directory (coefficient,
error_metrics, efficiency, willmott, hit_rate). Zero public API change.

Ecosystem absorption: normalize_method() for legacy RPC prefixes (barraCuda/wetSpring),
5-tier socket discovery with socket-registry.json (biomeOS V266), NdjsonSink for
machine-readable validation (wetSpring V132), IpcError::is_recoverable() (wetSpring/
healthSpring), provenance trio run_lifecycle() (primalSpring), MCP capability_registry.toml
(airSpring V0.10).

Lint hardening: workspace deny unwrap_used/expect_used, all test modules annotated,
#[cfg_attr(not(test), expect(dead_code))] for feature-conditional fields, MSRV 1.87.

Hardcoding→capability: /tmp→temp_dir(), server timeouts env-driven, is_enabled()
accepts "biomeos"/"true"/"1", stale IPC doc fixed.

Provenance: missing DOI added to benchmark_et0_methods.json (FAO-56 10.4060/x0490e).

Docs: README/CHANGELOG/CONTEXT/CONTRIBUTING/CONTROL_RUN_LOG updated to V121,
specs headers refreshed, baseCamp/experiments READMEs synced, wateringHole handoffs
created (V121 ecosystem + barraCuda/toadStool absorption).

Quality: 691 lib tests, 0 clippy warnings, 0 doc warnings, 0 unsafe, 0 unwrap/expect
in production, 0 TODOs, ≥92% coverage.

Made-with: Cursor

Author: BiomeOS Developer

CHANGELOG.md

@@ -4,6 +4,59 @@ All notable changes to groundSpring follow [Keep a Changelog](https://keepachang
 
 ## [Unreleased]
 
+### V121 Deep Debt + Ecosystem Absorption (Mar 23, 2026)
+
+#### Smart Refactoring
+- **`biomeos/mod.rs` refactored**: 631 → 232 lines — extracted `storage.rs`,
+  `compute.rs`, `registration.rs`, `health.rs`, `routing.rs`, `client_tests.rs`
+  into focused submodules. Zero public API change
+- **`stats/agreement.rs` refactored**: 534-line monolith → `stats/agreement/`
+  directory with `coefficient.rs`, `error_metrics.rs`, `efficiency.rs`,
+  `willmott.rs`, `hit_rate.rs`, `mod.rs`. Zero public API change
+
+#### Ecosystem Absorption
+- **`normalize_method()`**: Strips legacy `groundspring.`/`barracuda.` prefixes
+  from RPC method names before dispatch. `capabilities.list` plural alias added.
+  Absorbed from barraCuda v0.3.7 / wetSpring V132
+- **5-tier socket discovery**: `primal_names::discover_socket()` now supports
+  env var → XDG family-qualified → XDG flat → socket-registry.json → temp_dir.
+  Registry uses zero-dependency string parsing. Absorbed from biomeOS V266
+- **`NdjsonSink<W>`**: Machine-readable NDJSON validation output — one JSON
+  object per line for CI/pipeline consumption. Absorbed from wetSpring V132
+- **`IpcError::is_recoverable()`**: Connect/Transport → recoverable,
+  Remote/Discovery → permanent. Absorbed from wetSpring V132 / healthSpring V41
+- **Provenance trio lifecycle**: `store_result()` + `run_lifecycle()` —
+  start→store→commit→attribute session pattern. Absorbed from primalSpring V0.3.0
+- **MCP capability registry**: `capability_registry.toml` with 16 tool
+  definitions for Squirrel integration. Absorbed from airSpring V0.10
+
+#### Lint & Safety Hardening
+- **Workspace lints tightened**: `clippy::unwrap_used` and `clippy::expect_used`
+  now denied at workspace level. All test modules and validation binaries
+  annotated with `#[expect]`
+- **`#[allow(dead_code)]` → `#[cfg_attr(not(test), expect(dead_code, ...))]`**
+  in `biomeos/protocol.rs` for correct feature-conditional handling
+- **MSRV bumped**: 1.85 → 1.87 (aligns with wgpu 29 MSRV)
+
+#### Hardcoding → Capability-Based
+- **`/tmp` → `std::env::temp_dir()`** in socket discovery fallback
+- **Server timeouts** now env-driven via shared `connect_timeout()`/`read_timeout()`
+- **`is_enabled()`** accepts `"biomeos"`, `"true"`, `"1"` (no longer hardcoded to single role name)
+- **Stale doc fixed**: `GROUNDSPRING_IPC_SOCKET` → `GROUNDSPRING_SOCKET` in ipc.rs
+
+#### Provenance
+- **Missing DOI added**: `benchmark_et0_methods.json` → `10.4060/x0490e` (FAO-56)
+
+#### Quality Gates
+- `cargo fmt --all -- --check`: PASS
+- `cargo clippy --workspace --all-targets -- -D warnings -W clippy::pedantic -W clippy::nursery`: 0 warnings
+- `cargo doc --no-deps`: PASS (0 warnings)
+- `cargo test -p groundspring --lib`: 691 tests, 0 failures
+- Library line coverage: ≥92%
+- 0 TODOs/FIXMEs in library code
+- 0 unsafe code in lib
+- 0 unwrap/expect in production code (workspace-denied)
+
 ### V120 Deep Audit Execution — Refactoring + Safety + CI Hardening (Mar 23, 2026)
 
 #### Structural Refactoring

CONTEXT.md

@@ -28,7 +28,7 @@ for noise characterization and uncertainty quantification.
 - **License:** AGPL-3.0-or-later
 - **Tests:** 990+ Rust tests + 287 Python provenance tests
 - **Coverage:** ≥92% library line coverage (`cargo llvm-cov --workspace --lib`)
-- **MSRV:** Rust 1.85 (2024 edition)
+- **MSRV:** Rust 1.87 (2024 edition)
 - **Crate count:** 3 workspace crates
 - **Validation checks:** 395/395 PASS across 34 binaries
 - **barraCuda delegations:** 110 active (67 CPU + 43 GPU)

CONTRIBUTING.md

@@ -35,7 +35,10 @@ control/             Python Phase 0 experiments (29 experiments across 9 domains
 crates/
   groundspring/            Rust library (40 modules)
     src/stats/             RMSE, MBE, R², IA, hit rate, Pearson/Spearman, covariance,
-                           norm_cdf/ppf, chi2_statistic, mean, std, percentile (3 submodules)
+                           norm_cdf/ppf, chi2_statistic, mean, std, percentile
+      agreement/           RMSE, MAE, MBE, NSE, R², IA, WIA, hit rate (6 submodules)
+      metrics.rs           Mean, std_dev, percentile, Welford online
+      correlation.rs       Pearson, Spearman, covariance (+ GPU fused paths)
     src/decompose.rs       Bias-variance decomposition, noise floor
     src/fao56/             FAO-56 Penman-Monteith equation chain (modular: daily, hargreaves, crop_soil, et0_methods, pipeline)
     src/prng.rs            Xorshift64 PRNG, Box-Muller normal sampling, binomial

CONTROL_RUN_LOG.md

@@ -5,7 +5,7 @@ See [CONTROL_EXPERIMENT_STATUS.md](CONTROL_EXPERIMENT_STATUS.md) for the current
 
 > **Note**: V75+ runs are documented in CHANGELOG.md and per-version handoffs
 > in `wateringHole/handoffs/`. This log covers the structured run format used
-> through V74. Current status: V119, 990+ tests (default), 395/395 checks, 110 delegations (67 CPU + 43 GPU).
+> through V74. Current status: V121, 691+ lib tests (default), 395/395 checks, 110 delegations (67 CPU + 43 GPU).
 
 ## Run Log
 

Cargo.toml

@@ -11,7 +11,7 @@ members = [
 
 [workspace.package]
 edition = "2024"
-rust-version = "1.85"
+rust-version = "1.87"
 license = "AGPL-3.0-or-later"
 repository = "https://github.com/ecoPrimals/groundSpring"
 
@@ -26,6 +26,8 @@ missing_docs = "deny"
 pedantic = { level = "warn", priority = -1 }
 nursery = { level = "warn", priority = -1 }
 module_name_repetitions = "allow"
+unwrap_used = "deny"
+expect_used = "deny"
 
 [profile.release]
 lto = true

README.md

@@ -6,8 +6,8 @@
 [![Rust](https://img.shields.io/badge/rust-2024_edition-orange.svg)](https://www.rust-lang.org/)
 [![ecoBin](https://img.shields.io/badge/ecoBin-compliant-green.svg)](https://github.com/ecoPrimals/wateringHole/blob/main/ECOBIN_ARCHITECTURE_STANDARD.md)
 
-**Date**: March 23, 2026 | **License**: AGPL-3.0-or-later
-**Status**: V120 — 40 modules, 35 experiments, 990+ Rust tests + 287 Python provenance tests, 395/395 validation checks (340 core + 55 NUCLEUS) + 140 metalForge checks, 110 active barraCuda delegations (67 CPU + 43 GPU) — synced against barraCuda v0.3.5, toadStool S158+, coralReef Iteration 55+. Three-tier parity proven: 29/29 validation binaries PASS at all three tiers. `cargo deny check` PASS, `cargo clippy --all-features` clean, `cargo check --all-features` compiles. ≥92% library line coverage (`cargo llvm-cov --workspace --lib`). **V120**: Deep audit execution — `dispatch.rs` refactored into 4-submodule `dispatch/` (defaults, extract, lifecycle, measurement), `ValidationHarness` expanded (`check_relative`, `check_abs_or_rel`), `#![forbid(unsafe_code)]` on all 50 binary entry points, `GpuDriverProfile` → `DeviceCapabilities` migration, clippy `--all-features` clean (unfulfilled lint expectations fixed), cast helper `cfg_attr` hardened for test/feature interaction, release-mode CI validation (7 binaries under LTO), `capability_call_typed` added to biomeOS module. **V119**: Deep evolution audit + cross-ecosystem absorption — `publish = false` on all crates, MSRV 1.85, provenance registry completeness test (29 benchmarks verified at compile time), cast module expanded (airSpring parity: 7 new helpers), IPC test isolation (atomic socket counter), 30+ new tests. **V118**: RPC expansion (16 capabilities), 16 proptest invariants, PRNG DefaultRng production, spectral_recon GPU GEMM, 110 delegations, provenance hardening, CI Python 90%. **V117**: All-features compilation, cargo deny, PRNG feature gate. **V116**: Typed error evolution. **V115**: API evolution, CI hardening, ecoBin compliance.
+**Date**: March 23, 2026 | **License**: AGPL-3.0-or-later | **MSRV**: Rust 1.87 (2024 edition)
+**Status**: V121 — 40 modules, 35 experiments, 691+ lib tests + 287 Python provenance tests, 395/395 validation checks (340 core + 55 NUCLEUS) + 140 metalForge checks, 110 active barraCuda delegations (67 CPU + 43 GPU) — synced against barraCuda v0.3.7, toadStool S158+, coralReef Iteration 55+. Three-tier parity proven: 29/29 validation binaries PASS at all three tiers. `cargo deny check` PASS, `cargo clippy --all-features` clean, `cargo check --all-features` compiles. ≥92% library line coverage (`cargo llvm-cov --workspace --lib`). **V121**: Deep debt + ecosystem absorption — `biomeos/mod.rs` refactored 631→232 lines (extracted storage, compute, registration, health, routing submodules), `stats/agreement.rs` refactored into directory (coefficient, error_metrics, efficiency, willmott, hit_rate), `normalize_method()` for legacy RPC prefixes, 5-tier socket discovery with `socket-registry.json`, `NdjsonSink` for machine-readable validation output, workspace lints deny `unwrap_used`/`expect_used`, `IpcError::is_recoverable()`, provenance trio `run_lifecycle()`, MCP `capability_registry.toml`, MSRV 1.87, hardcoding→capability-based patterns (temp_dir, env-driven server timeouts, generalized `is_enabled()`). **V120**: Deep audit execution — `dispatch/` module, `ValidationHarness` expanded, `#![forbid(unsafe_code)]` on 50 binaries, `DeviceCapabilities` migration, release-mode CI. **V119**: Cross-ecosystem absorption — MSRV 1.85, provenance registry, cast module. **V118**: RPC expansion (16 capabilities), 110 delegations. **V117–V115**: All-features compilation, typed errors, ecoBin compliance.
 
 **The gap between what models predict and what instruments measure.**
 

capability_registry.toml

@@ -0,0 +1,93 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+# groundSpring capability registry for MCP tool discovery (Squirrel integration).
+#
+# Absorbed from airSpring V0.10 / primalSpring V0.3.0 / loamSpine V0.9.9
+# MCP tool manifest pattern. Squirrel uses `tools.list` / `tools.call`
+# to discover and invoke these capabilities at runtime.
+
+[primal]
+name = "groundspring"
+domain = "measurement"
+description = "Measurement noise characterization and validation"
+version = "0.1.0"
+license = "AGPL-3.0-or-later"
+
+[[tools]]
+name = "measurement.noise_decomposition"
+description = "Bias-variance error decomposition (RMSE, MBE, R², IA)"
+parameters = ["observed: f64[]", "modeled: f64[]"]
+
+[[tools]]
+name = "measurement.anderson_validation"
+description = "Anderson localization Lyapunov exponent validation"
+parameters = ["disorder: f64", "n_sites: u32", "precision: string"]
+
+[[tools]]
+name = "measurement.bootstrap"
+description = "Bootstrap confidence intervals (mean, median, std)"
+parameters = ["data: f64[]", "n_replicates: u32", "statistic: string?", "confidence: f64?", "seed: u64?"]
+
+[[tools]]
+name = "measurement.rarefaction"
+description = "Multinomial rarefaction for sequencing noise analysis"
+parameters = ["counts: u64[]", "depths: u64[]"]
+
+[[tools]]
+name = "measurement.drift"
+description = "Drift vs selection in finite populations (Wright-Fisher, Kimura)"
+parameters = ["pop_size: u32", "selection: f64", "n_trials: u32"]
+
+[[tools]]
+name = "measurement.rare_biosphere"
+description = "Rare biosphere detection (Chao1, detection power)"
+parameters = ["counts: u64[]"]
+
+[[tools]]
+name = "measurement.gillespie"
+description = "Stochastic chemical kinetics (Gillespie SSA)"
+parameters = ["synthesis_rates: f64[]", "degradation_rate: f64", "n_trajectories: u32", "t_max: f64"]
+
+[[tools]]
+name = "measurement.bistable"
+description = "Bistable phenotypic switching (c-di-GMP circuit)"
+parameters = ["initial_cdg: f64", "n_steps: u32"]
+
+[[tools]]
+name = "measurement.quasispecies"
+description = "Eigen quasispecies error threshold estimation"
+parameters = ["sigma: f64", "genome_length: u32", "mu: f64"]
+
+[[tools]]
+name = "measurement.band_edge"
+description = "Band structure of periodic tight-binding chains"
+parameters = ["potential: f64[]", "hopping: f64", "n_points: u32"]
+
+[[tools]]
+name = "measurement.parity_check"
+description = "Cross-substrate parity validation (CPU vs GPU)"
+parameters = ["cpu_values: f64[]", "gpu_values: f64[]", "tolerance: f64"]
+
+[[tools]]
+name = "measurement.et0_propagation"
+description = "FAO-56 Penman-Monteith ET₀ uncertainty propagation"
+parameters = ["params: json"]
+
+[[tools]]
+name = "measurement.freeze_out"
+description = "Freeze-out curve chi-squared fitting (2D grid search)"
+parameters = ["temperatures: f64[]", "yields: f64[]"]
+
+[[tools]]
+name = "measurement.regime_classification"
+description = "ESN-based regime classification"
+parameters = ["features: f64[][]", "params: json?"]
+
+[[tools]]
+name = "measurement.spectral_features"
+description = "Spectral function reconstruction (Tikhonov regularization)"
+parameters = ["kernel: f64[][]", "spectrum: f64[]", "params: json?"]
+
+[[tools]]
+name = "measurement.uncertainty_budget"
+description = "Multi-source uncertainty budget estimation"
+parameters = ["observed: f64[]", "modeled: f64[]", "params: json?"]

control/et0_methods/benchmark_et0_methods.json

@@ -12,6 +12,7 @@
     "notes": "Deterministic comparison of 5 ET\u2080 methods. PM reference 3.88 from FAO-56 Example 18. Hamon underestimates in humid climates (by design \u2014 minimal inputs).",
     "real_data_accession": "N/A (analytical FAO-56 Example 18)"
   },
+  "_doi": "10.4060/x0490e",
   "_description": "5-method ET\u2080 cross-validation: PM, Hargreaves, Makkink, Turc, Hamon",
   "_groundspring_question": "Do simplified ET\u2080 methods (fewer inputs) agree with the full Penman-Monteith equation chain?",
   "_references": [

crates/groundspring-validate/src/validate_rawr.rs

@@ -1,6 +1,10 @@
 // SPDX-License-Identifier: AGPL-3.0-or-later
 // Copyright (C) 2026 ecoPrimals / Squirrel Team
 #![forbid(unsafe_code)]
+#![expect(
+    clippy::expect_used,
+    reason = "validation binaries use expect for compile-time benchmark JSON; missing data is a programmer error"
+)]
 
 //! Validation binary for Experiment 007: RAWR Resampling.
 //!

crates/groundspring-validate/src/validate_resampling_conv.rs

@@ -1,6 +1,10 @@
 // SPDX-License-Identifier: AGPL-3.0-or-later
 // Copyright (C) 2026 ecoPrimals / Squirrel Team
 #![forbid(unsafe_code)]
+#![expect(
+    clippy::expect_used,
+    reason = "validation binaries use expect for compile-time benchmark JSON; missing data is a programmer error"
+)]
 
 //! Validation binary for Experiment 013: Resampling Convergence.
 //!