docs: v1.146.0 — x-FuSa spec v1.15.2, §1.6.1 Rule A false-positive tradeoff

CHANGELOG.md

@@ -7,6 +7,16 @@ and the project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.
 
 ## [Unreleased]
 
+## [1.146.0] — 2026-07-29
+
+### docs: x-FuSa spec v1.15.2 — §1.6.1 Rule A false-positive tradeoff made explicit (SoundMatt/FuSaOps#101)
+
+A same-day ada-FuSa self-review flagged Rule A's deny-list as occasionally
+false-positiving on legitimate prose (`buffer[i]`, `"STBD"`) and asked if
+that was intended. It is — confirmed and made explicit in the spec text with
+both examples, so future audits hit the documented answer instead of
+re-deriving it. Pure documentation addition, no behavior change.
+
 ## [1.145.0] — 2026-07-29
 
 ### docs: x-FuSa spec v1.15.1 — schemaVersion/specVersion clarified as MAJOR.MINOR.PATCH (SoundMatt/FuSaOps#99)

docs/x-fusa-spec.md

@@ -1,6 +1,6 @@
 # x-FuSa Tool Specification
 
-**Spec version:** 1.15.1 · **Status:** Normative · **Owner:** FuSaOps
+**Spec version:** 1.15.2 · **Status:** Normative · **Owner:** FuSaOps
 
 This is the **master contract** every x-FuSa tool (go-FuSa, c-FuSa, cpp-FuSa, and
 future tools) implements. It defines the CLI surface, the machine-readable output
@@ -502,6 +502,15 @@ itself still belongs to the artifact's own command.
   placeholder text real; a legitimate hazard/failure-mode description that
   happens to contain a deny-listed string is rare enough to warrant an
   explicit, individually-justified waiver rather than a blanket pass.
+  **This tradeoff is deliberate, not a bug to fix per-tool** — e.g. ordinary
+  engineering prose like `"buffer[i] causes memory corruption"` or the
+  abbreviation `"STBD"` (containing the `"TBD"` substring) will occasionally
+  false-positive against this canonical, byte-for-byte-identical-across-tools
+  deny-list; the fix for that specific finding is a disposition waiver
+  (`disposition add --reason "..."`), never a tool narrowing its own
+  detector's pattern to be "smarter" than the canonical one — two tools
+  disagreeing on what counts as a Rule A match undermines the "canonical"
+  guarantee this section exists to provide (filed as SoundMatt/FuSaOps#101).
 - **Rule B — blanket qualitative fallback (SHOULD, a WARNING by default).**
   For an artifact with **≥10 entries**, a tool SHOULD compute each
   qualitative field's **distinct-value ratio** (count of distinct values ÷
@@ -1815,6 +1824,24 @@ tool-defined row.
 
 ## 14. Changelog
 
+### 1.15.2 — 2026-07-29 (§1.6.1 Rule A: explicit false-positive example, filed as SoundMatt/FuSaOps#101)
+
+A same-day ada-FuSa self-review flagged Rule A's bracket/`"TBD"`-substring
+deny-list as producing occasional false positives on legitimate engineering
+prose (`"buffer[i] causes memory corruption"`, the abbreviation `"STBD"`) and
+asked whether this was an intended tradeoff or a spec gap. Re-reading the
+existing text confirmed it's intended — the deny-list is deliberately simple
+and canonical (byte-for-byte identical across tools) rather than
+word-boundary-aware, with occasional false positives meant to be resolved via
+an individual disposition waiver, not per-tool detector narrowing. The
+reviewer correctly did not narrow ada-FuSa's own detector on exactly that
+reasoning. Since the tradeoff was "evidently non-obvious enough that outside
+audit tooling flags it as a bug on first read" (the reviewer's own words),
+Rule A's definition now states the tradeoff explicitly with the same two
+examples, so a future implementer hits the documented answer instead of
+re-deriving it. Pure documentation addition, no behavior change — hence a
+PATCH bump.
+
 ### 1.15.1 — 2026-07-29 (schemaVersion/specVersion format clarified to MAJOR.MINOR.PATCH, filed as SoundMatt/FuSaOps#99)
 
 §3.2/§12's normative text said `schemaVersion` (and, by extension, `specVersion`)

fmea.json

@@ -1,11 +1,11 @@
 {
-  "schemaVersion": "1.15.1",
+  "schemaVersion": "1.15.2",
   "kind": "fmea-report",
   "language": "go",
-  "generatedAt": "2026-07-29T02:45:59.121719Z",
+  "generatedAt": "2026-07-29T02:58:10.51284Z",
   "projectRoot": ".",
   "tool": "fusaops",
-  "toolVersion": "1.145.0",
+  "toolVersion": "1.146.0",
   "standard": "iso26262",
   "ratingScale": "custom-1-10",
   "entries": [
@@ -210,5 +210,5 @@
     "coveragePct": 19.5,
     "componentInventoryMethod": "count of all packages in CLAUDE.md's core orchestration-pipeline table (11) and layered compliance/evidence/workflow table (30), 41 total; componentsAnalyzed counts distinct FMEA entries, not distinct packages named (some entries span multiple packages, e.g. \"auditpack / sign packages\")"
   },
-  "hash": "sha256:ad57ee1c0abac5b9244b27d3df0301bd9a64f2a0089764fe526bb027a7fb9f31"
+  "hash": "sha256:fe07227fc6654ce31da551365e1fefc10dabd6cba5008e2f072ea5e90cb079f9"
 }

fusaops.go

@@ -20,12 +20,12 @@ import (
 )
 
 // Version is the current release of FuSaOps.
-const Version = "1.145.0"
+const Version = "1.146.0"
 
 // SpecVersion is the x-FuSa specification version this release targets.
 //
 //fusa:req REQ-FO-CORE007
-const SpecVersion = "1.15.1"
+const SpecVersion = "1.15.2"
 
 // Sentinel errors. Callers should use errors.Is for comparison.
 //