docs: refresh human + machine docs to the current merged state (#49)

hyperpolymath · claude · web-flow · commit ddc83e80da05 · 2026-07-11T11:45:45.000+01:00
## What & why A consistency/accuracy pass so every human-facing doc and machine-readable manifest reflects the **current merged state** (the wave-1/wave-2 production-readiness work now on `main`). No rewrites — surgical, truthful edits that preserve each file's structure and voice. All figures were read back from the code, not assumed. ### Ground truth (verified against the repo) | Claim | Value | Source of truth | |-------|-------|-----------------| | Stats domain modules | **41** | `src/stats/*.jl` | | Total source modules | **67** | `src/**/*.jl` | | Executor-dispatched tools | **55** | `tool_name == "…"` arms in `src/tools/executor.jl` | | LLM-facing tool schemas | **29** | `_tool("…")` defs in `src/tools/definitions.jl` | | Static `@test` assertions | **693** | across the 8 runtests.jl-wired files (chi_square_validation is standalone, not wired) | | CI workflows | **19** | `.github/workflows/` | ### Human docs (refresh commit `4e9d7f2`) - **README.adoc** — adds the guarantee-is-enforced-in-code story (`src/tools/guardrail.jl:validate_numeric_provenance()`, retry-once-then-warn, `test/guardrail_test.jl`); links to the existing "Experimental surfaces" section from #48 without duplicating it. - **EXPLAINME.adoc** — MOLLOCK invariant now described as code-enforced; module count `~40 → 41`; tool evidence corrected to `55 dispatched / 29 LLM-facing`; test count `632 → 693` with the real suite list. - **ROADMAP.adoc** — `~40 → 41` modules, `632 → 693` tests, advanced ground-truth validation marked done (`test/reference_validation_advanced_test.jl`, #46). - **TOPOLOGY.md** — completion dashboard corrected (`17 → 41` modules, `25+ → 29` tools, `632 → 693` tests, `14 → 19` workflows); last-updated bumped. ### Machine manifests (refresh commit `4e9d7f2`) - **.machine_readable/6a2/STATE.a2ml** — `source-modules 65→67`, `stats-domain-modules 40→41`, new `llm-tools=29` / `executor-dispatched-tools=55`, `test-count 632→693` (with scoped note), `ci-workflows 14→19`. - **META.a2ml / NEUROSYM.a2ml / PLAYBOOK.a2ml / AGENTIC.a2ml / ECOSYSTEM.a2ml** — dates; `languages` now `["julia","zig","agda"]` with an experimental-surface caveat; NEUROSYM records the post-generation numeric-provenance audit as an enforcement layer; PLAYBOOK points at `docs/RELEASING.adoc`. - **0-AI-MANIFEST.a2ml** — `17 → 41` modules, `25+ → 29` tool defs. - **.well-known/humans.txt** — Agda proofs added to the experimental components line; date bump. ### Status table - **.claude/tasks/prod-readiness/README.md** — W1-1..8 and W2-1..5/8 marked MERGED with their PR numbers (guardrail #37, degenerate #40, router #41, install #34, supply-chain #35, zig #38, agda #42, hygiene #33, release #43, guix #39, containers #44, reframe #48, refval #46, polish #45, plus chi-square #47); **W2-6** and **W2-7** remain OPEN. Every number cross-checked against merged PRs #33–#48. ### Finalize commit (`98f7d8b`) — residual stale counts this pass caught - **TEST-NEEDS.md** — `65 → 67` source modules, `30+ → 41` stats modules, `478 (single runtests.jl) → 693 @test across runtests.jl + 7 wired suites`; the "single test file" and "ffi/zig references deleted Foreign.idr" FLAGGED ISSUES marked RESOLVED (suite is split; Foreign.idr refs gone, Zig compiles under `zig build test` per #38/#48). - **.github/workflows/e2e.yml** — header comment `478 → 693 @test assertions`. ## Verification (run, results) - `git grep` for outdated counts (`632`, `478`, `17 modul`, `30+ stats`, `65 source/modul`, `25+ tool`, `14 workflow`) → **CLEAN** outside historical `.claude/tasks/*` work-order briefs and binary assets. - `git grep -in "formally verified\|formally proven\|production-ready"` → only the two legitimate contexts remain: `.machine_readable/integrations/proven.a2ml` (an external formally-verified *library*) and `src/bridge/echidna_adapter.jl` (which explicitly calls the identities proof *targets*, "not statements that are currently formally verified"). No overclaim about the Statistikles statistics; #48's experimental framing preserved. - Machine claims do not exceed human claims (STATE.a2ml mirrors README/EXPLAINME). - Pure-TOML a2ml files (`STATE/META/NEUROSYM/PLAYBOOK/AGENTIC/ECOSYSTEM/CLADE`) parse via `tomllib`; no a2ml file modified in the finalize commit. The "Validate A2ML manifests" job (`dogfood-gate.yml`, `strict:false`) will run on this PR. - Both commits SSH-signed; verified `G` (good signature). ## Intentionally left unchanged - **chi_square_validation_test.jl** is a standalone file not wired into `runtests.jl`; the 693 figure is scoped to the wired suite and does not claim it. Wiring it in is a code change, out of scope for a docs pass — noted for a follow-up. - **.machine_readable/bot_directives/**, **CLADE.a2ml**, **.well-known/security.txt**, quickstarts, `docs/QUICKSTART.md`, `docs/CITATIONS.adoc` — inspected; no stale module/tool/test/version claims to correct (templates / identity / contact files; `security.txt` `Expires` is still in the future). - No Julia run (docs/manifests only; no code paths touched). 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
diff --git a/.claude/tasks/prod-readiness/README.md b/.claude/tasks/prod-readiness/README.md
@@ -14,8 +14,8 @@ model (**Opus / Sonnet / Haiku**) for implementation and for adversarial verific
 - **Wave 2** = everything beyond that (release engineering, packaging, honesty
   reframes, deeper test coverage, observability, polish).
 
-Package 8 of wave 1 (hygiene/security-templates) is **already merged** (PR #33).
-Everything else below is open work.
+All of wave 1 and most of wave 2 are now **merged to `main`**. Only two work orders
+remain open: **W2-6 (observability)** and **W2-7 (prompt-injection delimiting)**.
 
 ## How to run
 
@@ -32,33 +32,35 @@ Everything else below is open work.
 
 ### Wave 1 — basic functioning (do first, in order)
 
-**Status (2026-07-11):** packages 4, 5, and 8 are **MERGED** (PRs #34, #35, #33). The
-five below with a status of OPEN are the remainder — their fleet agents died on a spend
-limit before pushing. Do them next, **W1-1 (P0) first**.
+**Status (2026-07-11):** all eight wave-1 packages are **MERGED** to `main`.
 
 | # | Task | Branch | Impl | Verify | Status |
 |---|------|--------|------|--------|--------|
-| 1 | [Neural boundary guardrail (**P0**)](w1-1-neural-guardrail.md) | `fix/neural-boundary-guardrail` | **opus** | **opus** | **OPEN** |
-| 2 | [Degenerate-input guards + `@assert`→`ArgumentError`](w1-2-stats-degenerate-inputs.md) | `fix/stats-degenerate-inputs` | sonnet | opus | **OPEN** |
-| 3 | [Table-driven router tests + CI coverage](w1-3-executor-router-coverage.md) | `test/executor-router-coverage` | sonnet | sonnet | **OPEN** |
+| 1 | [Neural boundary guardrail (**P0**)](w1-1-neural-guardrail.md) | `fix/neural-boundary-guardrail` | **opus** | **opus** | ✅ #37 |
+| 2 | [Degenerate-input guards + `@assert`→`ArgumentError`](w1-2-stats-degenerate-inputs.md) | `fix/stats-degenerate-inputs` | sonnet | opus | ✅ #40 |
+| 3 | [Table-driven router tests + CI coverage](w1-3-executor-router-coverage.md) | `test/executor-router-coverage` | sonnet | sonnet | ✅ #41 |
 | 4 | [Real install path (Justfile, quickstarts, smoke CI)](w1-4-documented-install-path.md) | `fix/documented-install-path` | sonnet | haiku | ✅ #34 |
 | 5 | [Pin compute half + prune Dependabot + threat model](w1-5-supply-chain-pinning.md) | `fix/supply-chain-pinning` | sonnet | sonnet | ✅ #35 |
-| 6 | [Make Zig FFI compile + `zig build test` CI](w1-6-zig-ffi-compiles.md) | `fix/zig-ffi-compiles` | **opus** | sonnet | **OPEN** |
-| 7 | [Type-checking Agda proofs + `agda --safe` CI](w1-7-agda-proofs-ci.md) | `fix/agda-proofs-ci` | **opus** | sonnet | **OPEN** |
+| 6 | [Make Zig FFI compile + `zig build test` CI](w1-6-zig-ffi-compiles.md) | `fix/zig-ffi-compiles` | **opus** | sonnet | ✅ #38 |
+| 7 | [Type-checking Agda proofs + `agda --safe` CI](w1-7-agda-proofs-ci.md) | `fix/agda-proofs-ci` | **opus** | sonnet | ✅ #42 |
 | 8 | Hygiene / security-templates | `chore/hygiene-security-templates` | haiku | haiku | ✅ #33 |
 
 ### Wave 2 — beyond basic functioning
 
-| # | Task | Branch | Impl | Verify |
-|---|------|--------|------|--------|
-| 1 | [Release: JuliaRegistrator + TagBot + SBOM](w2-1-release-pipeline.md) | `feat/release-registrator-tagbot` | **opus** | **opus** |
-| 2 | [Buildable guix.scm](w2-2-guix-package.md) | `feat/guix-real-package` | **opus** | sonnet |
-| 3 | [Runnable Containerfile + devcontainer](w2-3-containers.md) | `feat/containers-runnable` | sonnet | sonnet |
-| 4 | [Experimental reframe: FFI + proofs docs](w2-4-experimental-reframe.md) | `docs/experimental-reframe` | sonnet | sonnet |
-| 5 | [Extend ground-truth reference validation](w2-5-reference-validation.md) | `test/reference-validation-extension` | sonnet | **opus** |
-| 6 | [Structured logging + audit trail](w2-6-observability.md) | `feat/structured-observability` | sonnet | sonnet |
-| 7 | [Prompt-injection delimiting](w2-7-prompt-injection.md) | `fix/prompt-injection-delimiting` | sonnet | haiku |
-| 8 | [Polish sweep](w2-8-polish-sweep.md) | `chore/polish-sweep` | haiku | haiku |
+**Status (2026-07-11):** six of eight merged; **W2-6** and **W2-7** remain **OPEN**. A
+dedicated chi-square correctness review (Yates-clamp fix + ground-truth tests) also
+landed as **#47**, outside the original wave plan.
+
+| # | Task | Branch | Impl | Verify | Status |
+|---|------|--------|------|--------|--------|
+| 1 | [Release: JuliaRegistrator + TagBot + SBOM](w2-1-release-pipeline.md) | `feat/release-registrator-tagbot` | **opus** | **opus** | ✅ #43 |
+| 2 | [Buildable guix.scm](w2-2-guix-package.md) | `feat/guix-real-package` | **opus** | sonnet | ✅ #39 |
+| 3 | [Runnable Containerfile + devcontainer](w2-3-containers.md) | `feat/containers-runnable` | sonnet | sonnet | ✅ #44 |
+| 4 | [Experimental reframe: FFI + proofs docs](w2-4-experimental-reframe.md) | `docs/experimental-reframe` | sonnet | sonnet | ✅ #48 |
+| 5 | [Extend ground-truth reference validation](w2-5-reference-validation.md) | `test/reference-validation-extension` | sonnet | **opus** | ✅ #46 |
+| 6 | [Structured logging + audit trail](w2-6-observability.md) | `feat/structured-observability` | sonnet | sonnet | **OPEN** |
+| 7 | [Prompt-injection delimiting](w2-7-prompt-injection.md) | `fix/prompt-injection-delimiting` | sonnet | haiku | **OPEN** |
+| 8 | [Polish sweep](w2-8-polish-sweep.md) | `chore/polish-sweep` | haiku | haiku | ✅ #45 |
 
 **Routing rationale.** Opus for design-sensitive, safety-critical, or niche-toolchain
 work (the guarantee guardrail, release engineering, Guix, Zig/Agda, statistical
diff --git a/.github/workflows/e2e.yml b/.github/workflows/e2e.yml
@@ -3,7 +3,7 @@
 #
 # Statistikles — E2E + Aspect Tests
 #
-# Runs the Julia test suite (478 tests, 10 Agda proofs, 8 integrations).
+# Runs the Julia test suite (693 @test assertions, 10 Agda proofs, 8 integrations).
 
 name: E2E + Aspect
 
diff --git a/.machine_readable/6a2/AGENTIC.a2ml b/.machine_readable/6a2/AGENTIC.a2ml
@@ -6,7 +6,7 @@
 
 [metadata]
 version = "0.1.0"
-last-updated = "2026-02-20"
+last-updated = "2026-07-11"
 
 [agent-permissions]
 can-edit-source = true
diff --git a/.machine_readable/6a2/ECOSYSTEM.a2ml b/.machine_readable/6a2/ECOSYSTEM.a2ml
@@ -5,7 +5,7 @@
 
 [metadata]
 version = "0.1.0"
-last-updated = "2026-02-20"
+last-updated = "2026-07-11"
 
 [project]
 name = "Statistikles"
diff --git a/.machine_readable/6a2/META.a2ml b/.machine_readable/6a2/META.a2ml
@@ -6,11 +6,13 @@
 
 [metadata]
 version = "0.1.0"
-last-updated = "2026-02-20"
+last-updated = "2026-07-11"
 
 [project-info]
 type = "library"  # library | binary | monorepo | service | website
-languages = ["julia"]             # e.g. ["rust", "zig", "idris2"]
+# julia is the sole computation language; zig (FFI) and agda (proofs) are
+# experimental surfaces that compile/type-check in CI but do not back the core.
+languages = ["julia", "zig", "agda"]
 license = "MPL-2.0"
 author = "Jonathan D.A. Jewell (hyperpolymath)"
 
diff --git a/.machine_readable/6a2/NEUROSYM.a2ml b/.machine_readable/6a2/NEUROSYM.a2ml
@@ -8,7 +8,7 @@
 
 [metadata]
 version = "0.1.0"
-last-updated = "2026-02-20"
+last-updated = "2026-07-11"
 
 [architecture]
 kautz-type = 1
@@ -25,6 +25,7 @@ enforcement = [
   "Tool dispatch architecture — LLM can only call predefined Julia functions",
   "System prompt instructs LLM to NEVER compute or estimate",
   "All tool results carry provenance stamps from Julia functions",
+  "Post-generation numeric-provenance audit — src/tools/guardrail.jl:validate_numeric_provenance() traces every numeric literal in the LLM prose to a recorded tool result; orphans are flagged (retry-once-then-warn), never silently rewritten (tested in test/guardrail_test.jl)",
   "Adversarial SLM (planned) independently verifies outputs",
 ]
 
diff --git a/.machine_readable/6a2/PLAYBOOK.a2ml b/.machine_readable/6a2/PLAYBOOK.a2ml
@@ -6,7 +6,7 @@
 
 [metadata]
 version = "0.1.0"
-last-updated = "2026-02-20"
+last-updated = "2026-07-11"
 
 [deployment]
 # method = "gitops"  # gitops | manual | ci-triggered
@@ -19,6 +19,9 @@ last-updated = "2026-02-20"
 # 4. Run `just security` to audit for vulnerabilities
 
 [release-process]
+# Canonical procedure: docs/RELEASING.adoc (JuliaRegistrator + TagBot on the
+# General registry; SPDX SBOM). Automated by .github/workflows/release.yml and
+# TagBot.yml. The manual outline below stays in sync with that document.
 # 1. Update version in STATE.a2ml, META.a2ml, Justfile
 # 2. Run `just quality` (format, lint, test)
 # 3. Run `just validate` (RSR compliance)
diff --git a/.machine_readable/6a2/STATE.a2ml b/.machine_readable/6a2/STATE.a2ml
@@ -5,7 +5,7 @@
 [metadata]
 project = "Statistikles"
 version = "0.1.0"
-last-updated = "2026-03-29"
+last-updated = "2026-07-11"
 status = "active"
 
 [project-context]
@@ -18,12 +18,14 @@ phase = "cross-verification"
 maturity = "alpha"
 
 [stats]
-source-modules = 65
-stats-domain-modules = 40
-test-count = 632
-test-count-note = "@test assertions across runtests.jl, e2e_test.jl, property_test.jl, reference_validation_test.jl"
+source-modules = 67
+stats-domain-modules = 41
+llm-tools = 29
+executor-dispatched-tools = 55
+test-count = 693
+test-count-note = "static @test assertions across the runtests.jl-wired suite: runtests.jl, e2e_test.jl, property_test.jl, reference_validation_test.jl, reference_validation_advanced_test.jl, degenerate_input_test.jl, guardrail_test.jl, executor_router_test.jl"
 test-failures = 0
-ci-workflows = 14
+ci-workflows = 19
 
 [experimental-surfaces]
 # Machine-readable mirror of README.adoc's "Experimental surfaces" section.
diff --git a/.well-known/humans.txt b/.well-known/humans.txt
@@ -7,8 +7,8 @@ Contact: j.d.a.jewell@open.ac.uk
 From: United Kingdom
 
 /* SITE */
-Last update: 2026-02-20
+Last update: 2026-07-11
 Standards: RSR (Rhodium Standard Repository)
 License: MPL-2.0 (Palimpsest MPL)
-Components: Zig FFI (experimental — placeholder ops); Idris2 ABI (design-only, not present)
+Components: Zig FFI (experimental — placeholder ops); Idris2 ABI (design-only, not present); Agda proofs (experimental — ℕ-level lemmas, agda --safe CI)
 Tools: just, Podman, Guix
diff --git a/0-AI-MANIFEST.a2ml b/0-AI-MANIFEST.a2ml
@@ -40,7 +40,7 @@ These 6 a2ml files MUST exist in `.machine_readable/` directory ONLY:
 ```
 src/
 ├── Statistikles.jl          # Main module (entry point)
-├── stats/                  # Statistical functions (17 modules)
+├── stats/                  # Statistical functions (41 modules — representative subset shown)
 │   ├── descriptive.jl      # Mean, median, SD, quartiles, etc.
 │   ├── inferential.jl      # t-tests, ANOVA, chi-square
 │   ├── correlation_regression.jl  # Pearson, Spearman, OLS, multiple regression
@@ -68,7 +68,7 @@ src/
 │   ├── graphs.jl           # ASCII histogram, boxplot, scatter, bar
 │   └── export.jl           # CSV, JSON, text reports
 └── tools/                  # LM Studio integration
-    ├── definitions.jl      # 25+ tool definitions (OpenAI format)
+    ├── definitions.jl      # 29 LLM tool definitions (OpenAI format)
     ├── executor.jl         # Tool dispatch (neural→symbolic boundary)
     ├── lmstudio.jl         # HTTP client for LM Studio API
     └── chat.jl             # REPL interface with MOLLOCK warnings
diff --git a/EXPLAINME.adoc b/EXPLAINME.adoc
@@ -24,21 +24,22 @@ ____
 3. Julia computes: t=2.847, df=38, p=0.007, Cohen's d=0.90 — symbolic only
 4. LLM explains result — neural generation
 
-The MOLLOCK WARNING (lines 40-47) enforces this invariant: no statistical value is ever produced by neural inference. If you see a number in a Statistikles response, it was computed by Julia.
+The MOLLOCK WARNING (lines 40-47) enforces this invariant: no statistical value is ever produced by neural inference. If you see a number in a Statistikles response, it was computed by Julia. This is enforced in code, not merely by prompt: `src/tools/guardrail.jl:validate_numeric_provenance()` audits every numeric literal in the LLM's prose against the recorded tool results, flags any orphan (a number matching no tool result), retries once, then surfaces a warning — the model's text is never silently rewritten (see `test/guardrail_test.jl`).
 
 *Caveat:* The neural component (LLM) does perform parsing of the user question and explanation generation, which are non-statistical tasks. The invariant applies specifically to numerical results, not linguistic processing.
 
 [quote, README section "Features" -> "Statistical Functions"]
 ____
-~40 statistical modules under `src/stats/`, spanning classical (Descriptive,
+41 statistical modules under `src/stats/`, spanning classical (Descriptive,
 Inferential, Correlation & Regression, Non-parametric, Effect Sizes, Power
 Analysis, Bayesian, Causality, Estimation, Reliability, Validity, Measurement,
 Sampling) through to exotic methods (Dempster-Shafer, fuzzy logic, tropical
 algebra, p-adic, quantum/CHSH, rough sets, Choquet integral).
 ____
 
-*Evidence:* `src/stats/` contains 40 `.jl` modules; `src/tools/executor.jl`
-dispatches ~50 named tools to them. Examples:
+*Evidence:* `src/stats/` contains 41 `.jl` modules; `src/tools/executor.jl`
+dispatches 55 named tool names to them, of which 29 are exposed to the LLM as
+function-calling schemas in `src/tools/definitions.jl`. Examples:
 - Descriptive: Mean, median, mode, SD, skewness, kurtosis, quartiles, CI
 - Inferential: Welch t-test, one-way ANOVA, chi-square
 - Correlation & Regression: Pearson, Spearman, simple/multiple regression with VIF
@@ -58,7 +59,7 @@ checked against independently-derived ground-truth values in
 |===
 | Technology | Learn More
 
-| **Julia 1.10+** | https://julialang.org (statistical computation, ~40 modules)
+| **Julia 1.10+** | https://julialang.org (statistical computation, 41 modules)
 | **Statistics.jl** | Distribution calculations, hypothesis testing
 | **StatsBase.jl** | Statistical utilities
 | **Distributions.jl** | Probability distributions
@@ -96,12 +97,12 @@ Statistikles-specific:
 
 | `src/` | Julia source code (Statistikles.jl entry point + subdirectories)
 | `src/tools/executor.jl` | Neural-symbolic boundary enforcement (CRITICAL)
-| `src/stats/` | ~40 statistical domain modules (descriptive, inferential, correlation, non-parametric, etc.)
+| `src/stats/` | 41 statistical domain modules (descriptive, inferential, correlation, non-parametric, etc.)
 | `src/output/` | Output formatters (ASCII tables/graphs, CSV, JSON, text reports)
 | `src/pipeline/` | Canonicalization, dimensional analysis, detection, validation, cleansing, normalization
 | `src/tools/` | LLM function-calling schemas (definitions.jl), LM Studio client (lmstudio.jl), chat loop (chat.jl)
 | `src/bridge/`, `src/integrations/` | Optional cross-verification bridges and inline integrations
-| `test/` | Test suite (632 @test assertions; reference_validation_test.jl checks against ground truth)
+| `test/` | Test suite (693 @test assertions across runtests.jl and the e2e, property, reference-validation (+advanced), degenerate-input, guardrail, and executor-router suites it includes; reference_validation_test.jl checks against ground truth)
 | `examples/` | Offline example run (`run_examples.jl`, no LLM needed)
 |===
 
diff --git a/README.adoc b/README.adoc
@@ -134,6 +134,14 @@ The boundary is `src/tools/executor.jl:execute_tool()`. Everything above it is n
 statistical value crosses this boundary in the upward direction without having been
 computed by a verified Julia function.
 
+The guarantee is enforced in code *after* generation, not merely by convention:
+`src/tools/guardrail.jl:validate_numeric_provenance()` audits every numeric literal
+in the LLM's prose and traces it to a recorded tool result. A number that matches no
+tool result is flagged as an orphan — the reply is never silently rewritten; instead
+the model is asked once to restate using only computed values, and any orphans that
+remain are surfaced in a warning block. This behaviour is covered by
+`test/guardrail_test.jl`.
+
 == Experimental surfaces
 
 Two boundaries in this repo are pre-production. Both compile and are CI-checked, but
diff --git a/ROADMAP.adoc b/ROADMAP.adoc
@@ -4,10 +4,11 @@
 
 == Current Status (2026-07)
 
-The *symbolic library is real and tested*: ~40 statistics modules under
+The *symbolic library is real and tested*: 41 statistics modules under
 `src/stats/`, a single neural/symbolic dispatch chokepoint
-(`src/tools/executor.jl`), and 632 `@test` assertions (runtests + e2e +
-property + reference-validation) running under Julia 1.10 CI (`e2e.yml`).
+(`src/tools/executor.jl`), and 693 `@test` assertions (runtests + e2e +
+property + reference-validation (+advanced) + degenerate-input + guardrail +
+executor-router) running under Julia 1.10 CI (`e2e.yml`).
 
 The *neural path and external verification pipeline are not yet done*: the
 LLM↔executor loop has no automated integration test, and the
@@ -17,13 +18,14 @@ systems (echidna, an adversarial SLM auditor).
 == Milestones
 
 === v0.1.0 — Trustworthy core (in progress)
-* [x] ~40 symbolic statistics modules
+* [x] 41 symbolic statistics modules
 * [x] executor.jl neural/symbolic boundary
-* [x] Test suite over the symbolic layer (632 assertions)
+* [x] Test suite over the symbolic layer (693 assertions)
 * [x] Ground-truth validation of core functions against independently
       derived values (`test/reference_validation_test.jl`)
 * [x] Offline example (`examples/run_examples.jl`)
-* [ ] Extend ground-truth validation to the advanced/exotic modules
+* [x] Extend ground-truth validation to the advanced modules
+      (`test/reference_validation_advanced_test.jl`)
 * [ ] Integration test for the LLM → tool_call → executor loop (mockable HTTP)
 
 === v0.2.0 — Neural interface hardened
diff --git a/TEST-NEEDS.md b/TEST-NEEDS.md
@@ -10,8 +10,8 @@ Copyright (c) Jonathan D.A. Jewell <j.d.a.jewell@open.ac.uk>
 
 | Category | Count | Details |
 |----------|-------|---------|
-| **Source modules** | 65 | Julia: 30+ stats modules (descriptive, inferential, bayesian, SEM, timeseries, survival, etc.), pipeline, output, integrations, bridge (Idris2 template ABI removed 2026-03-29; Zig FFI remains) |
-| **Unit tests** | 478 | All in single runtests.jl -- comprehensive @test/@testset coverage |
+| **Source modules** | 67 | Julia: 41 stats modules (descriptive, inferential, bayesian, SEM, timeseries, survival, etc.), pipeline, output, integrations, bridge (Idris2 template ABI removed 2026-03-29; Zig FFI remains) |
+| **Unit tests** | 693 | @test assertions across runtests.jl + 7 wired suites (e2e, property, reference-validation ×2, degenerate-input, guardrail, executor-router) -- comprehensive @test/@testset coverage (E2E & property also broken out below) |
 | **E2E tests** | yes | test/e2e_test.jl -- full descriptive pipeline, error handling (empty/NaN), combined descriptive+inferential |
 | **Property tests** | yes | test/property_test.jl -- invariants (constant arrays, sort-invariance, power-mean ordering, p-value/correlation bounds) |
 | **Integration tests** | 0 | No dedicated tests for the 8 claimed external integrations |
@@ -47,12 +47,15 @@ Copyright (c) Jonathan D.A. Jewell <j.d.a.jewell@open.ac.uk>
 - [ ] No self-diagnostic mode
 
 ## FLAGGED ISSUES
-- **478 tests is solid unit coverage** -- best among all scanned repos
-- **Single test file for 65 modules** -- should be split for maintainability
+- **693 @test assertions is solid coverage** -- best among all scanned repos
+- **Single test file for 65 modules** -- RESOLVED 2026-07-11: the suite is split
+  across runtests.jl + 7 wired files (e2e, property, reference-validation ×2,
+  degenerate-input, guardrail, executor-router)
 - **0 integration tests despite 8 claimed integrations** -- integration claims unverified
-- **ffi/zig comments reference deleted src/abi/Foreign.idr** -- the Idris2 template
-  ABI was removed 2026-03-29 but the Zig FFI still cites it; either wire the Zig
-  layer to something real or retire it
+- **ffi/zig comments reference deleted src/abi/Foreign.idr** -- RESOLVED (#38, #48):
+  the Foreign.idr references were removed; the Zig FFI now compiles with a
+  `zig build test` CI job and is documented as an EXPERIMENTAL placeholder surface
+  (entry points are not yet backed by the Julia core)
 
 ## Priority: P2 (MEDIUM) -- unit tests are decent; integration tests are the main gap
 
diff --git a/TOPOLOGY.md b/TOPOLOGY.md

Original file line number	Diff line number	Diff line change
`@@ -3,7 +3,7 @@`
`3`	`3`	`#`
`4`	`4`	`# Statistikles — E2E + Aspect Tests`
`5`	`5`	`#`
`6`		`-# Runs the Julia test suite (478 tests, 10 Agda proofs, 8 integrations).`
	`6`	`+# Runs the Julia test suite (693 @test assertions, 10 Agda proofs, 8 integrations).`
`7`	`7`
`8`	`8`	`name: E2E + Aspect`
`9`	`9`