@sun-asterisk/sungen 3.2.16-beta.12 → 3.2.16-beta.13

package/dist/harness/viewpoint-baseline.d.ts

@@ -0,0 +1,49 @@
+/**
+ * Keep the yardstick from moving with the thing it measures.
+ *
+ * `test-viewpoint.md` is the project's declaration of WHAT must be tested, and two
+ * scored axes measure the suite against it: `traceability` (do scenarios use the
+ * declared viewpoint ids?) and `atomicLedger` (is every declared item covered?).
+ * Both are only evidence while the declaration is INDEPENDENT of the suite.
+ *
+ * A `/sungen:create-test` run rewrote a filled test-viewpoint.md, replacing the
+ * declared categories with exactly the ones it had just generated. Both axes read
+ * 100% by construction — and the performance viewpoint it dropped on the way
+ * (4 perf scenarios → 2) went unreported, because VP-PERF was no longer declared
+ * for anything to be missing from.
+ *
+ * So: remember the declaration, and when it changes, say so and stop counting the
+ * two axes that depend on it until a human accepts the new baseline. Same rule the
+ * score model already applies to an undetermined page type — absent independent
+ * evidence scores nothing, in either direction.
+ */
+/** Axes whose evidence is `test-viewpoint.md` — excluded while the baseline is unconfirmed. */
+export declare const VIEWPOINT_DEPENDENT_AXES: readonly ["atomicLedger", "traceability"];
+export interface ViewpointBaseline {
+    status: 'absent' | 'new' | 'unchanged' | 'changed';
+    /** sha1 of the normalized file — '' when the unit has no test-viewpoint.md. */
+    hash: string;
+    ids: string[];
+    /** Only on `changed`: how the declaration moved. */
+    added?: string[];
+    removed?: string[];
+    recordedAt?: string;
+}
+/**
+ * Compare the unit's current test-viewpoint declaration to the recorded baseline.
+ * Read-only — recording is the caller's decision, so an audit stays a measurement
+ * and `--accept-viewpoint` stays the explicit human act. A first sighting reports
+ * `new` (nothing to compare against yet).
+ */
+export declare function checkViewpointBaseline(projectRoot: string, unitId: string, viewpointPath: string): ViewpointBaseline;
+/** Record the current declaration as the accepted baseline (`sungen audit --accept-viewpoint`). */
+export declare function acceptViewpointBaseline(projectRoot: string, unitId: string, current: {
+    hash: string;
+    ids: string[];
+}): void;
+/** Read the declaration without recording anything (used by `--accept-viewpoint`). */
+export declare function readViewpointDeclaration(viewpointPath: string): {
+    hash: string;
+    ids: string[];
+} | null;
+//# sourceMappingURL=viewpoint-baseline.d.ts.map

package/dist/harness/viewpoint-baseline.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"viewpoint-baseline.d.ts","sourceRoot":"","sources":["../../src/harness/viewpoint-baseline.ts"],"names":[],"mappings":"AAKA;;;;;;;;;;;;;;;;;;GAkBG;AAEH,+FAA+F;AAC/F,eAAO,MAAM,wBAAwB,2CAA4C,CAAC;AAElF,MAAM,WAAW,iBAAiB;IAChC,MAAM,EAAE,QAAQ,GAAG,KAAK,GAAG,WAAW,GAAG,SAAS,CAAC;IACnD,+EAA+E;IAC/E,IAAI,EAAE,MAAM,CAAC;IACb,GAAG,EAAE,MAAM,EAAE,CAAC;IACd,oDAAoD;IACpD,KAAK,CAAC,EAAE,MAAM,EAAE,CAAC;IACjB,OAAO,CAAC,EAAE,MAAM,EAAE,CAAC;IACnB,UAAU,CAAC,EAAE,MAAM,CAAC;CACrB;AAoCD;;;;;GAKG;AACH,wBAAgB,sBAAsB,CACpC,WAAW,EAAE,MAAM,EACnB,MAAM,EAAE,MAAM,EACd,aAAa,EAAE,MAAM,GACpB,iBAAiB,CAqBnB;AAED,mGAAmG;AACnG,wBAAgB,uBAAuB,CACrC,WAAW,EAAE,MAAM,EACnB,MAAM,EAAE,MAAM,EACd,OAAO,EAAE;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,GAAG,EAAE,MAAM,EAAE,CAAA;CAAE,GACvC,IAAI,CAQN;AAED,sFAAsF;AACtF,wBAAgB,wBAAwB,CAAC,aAAa,EAAE,MAAM,GAAG;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,GAAG,EAAE,MAAM,EAAE,CAAA;CAAE,GAAG,IAAI,CAKtG"}

package/dist/harness/viewpoint-baseline.js

@@ -0,0 +1,141 @@
+"use strict";
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.VIEWPOINT_DEPENDENT_AXES = void 0;
+exports.checkViewpointBaseline = checkViewpointBaseline;
+exports.acceptViewpointBaseline = acceptViewpointBaseline;
+exports.readViewpointDeclaration = readViewpointDeclaration;
+const crypto = __importStar(require("crypto"));
+const fs = __importStar(require("fs"));
+const path = __importStar(require("path"));
+const parse_1 = require("./parse");
+/**
+ * Keep the yardstick from moving with the thing it measures.
+ *
+ * `test-viewpoint.md` is the project's declaration of WHAT must be tested, and two
+ * scored axes measure the suite against it: `traceability` (do scenarios use the
+ * declared viewpoint ids?) and `atomicLedger` (is every declared item covered?).
+ * Both are only evidence while the declaration is INDEPENDENT of the suite.
+ *
+ * A `/sungen:create-test` run rewrote a filled test-viewpoint.md, replacing the
+ * declared categories with exactly the ones it had just generated. Both axes read
+ * 100% by construction — and the performance viewpoint it dropped on the way
+ * (4 perf scenarios → 2) went unreported, because VP-PERF was no longer declared
+ * for anything to be missing from.
+ *
+ * So: remember the declaration, and when it changes, say so and stop counting the
+ * two axes that depend on it until a human accepts the new baseline. Same rule the
+ * score model already applies to an undetermined page type — absent independent
+ * evidence scores nothing, in either direction.
+ */
+/** Axes whose evidence is `test-viewpoint.md` — excluded while the baseline is unconfirmed. */
+exports.VIEWPOINT_DEPENDENT_AXES = ['atomicLedger', 'traceability'];
+function baselinePath(projectRoot) {
+    return path.join(projectRoot, '.sungen', 'viewpoint-baseline.json');
+}
+function readBaselineFile(projectRoot) {
+    const p = baselinePath(projectRoot);
+    if (!fs.existsSync(p))
+        return { version: 1, units: {} };
+    try {
+        const parsed = JSON.parse(fs.readFileSync(p, 'utf-8'));
+        return parsed && typeof parsed === 'object' && parsed.units ? parsed : { version: 1, units: {} };
+    }
+    catch {
+        return { version: 1, units: {} }; // unreadable → treat as unrecorded, never throw
+    }
+}
+function writeBaselineFile(projectRoot, data) {
+    const p = baselinePath(projectRoot);
+    fs.mkdirSync(path.dirname(p), { recursive: true });
+    fs.writeFileSync(p, `${JSON.stringify(data, null, 2)}\n`, 'utf-8');
+}
+/**
+ * Hash the DECLARATION, not the prose: an author reflowing a sentence must not read
+ * as the taxonomy moving. Only the declared viewpoint ids are fingerprinted.
+ */
+function fingerprint(ids) {
+    return crypto.createHash('sha1').update(ids.join('\n')).digest('hex').slice(0, 12);
+}
+/**
+ * Compare the unit's current test-viewpoint declaration to the recorded baseline.
+ * Read-only — recording is the caller's decision, so an audit stays a measurement
+ * and `--accept-viewpoint` stays the explicit human act. A first sighting reports
+ * `new` (nothing to compare against yet).
+ */
+function checkViewpointBaseline(projectRoot, unitId, viewpointPath) {
+    if (!fs.existsSync(viewpointPath))
+        return { status: 'absent', hash: '', ids: [] };
+    const ids = (0, parse_1.parseViewpointOverview)(viewpointPath).map((v) => v.id).sort();
+    const hash = fingerprint(ids);
+    if (ids.length === 0)
+        return { status: 'absent', hash: '', ids: [] };
+    const file = readBaselineFile(projectRoot);
+    const recorded = file.units[unitId];
+    if (!recorded)
+        return { status: 'new', hash, ids };
+    if (recorded.hash === hash)
+        return { status: 'unchanged', hash, ids, recordedAt: recorded.recordedAt };
+    const before = new Set(recorded.ids ?? []);
+    const now = new Set(ids);
+    return {
+        status: 'changed',
+        hash,
+        ids,
+        added: ids.filter((i) => !before.has(i)),
+        removed: (recorded.ids ?? []).filter((i) => !now.has(i)),
+        recordedAt: recorded.recordedAt,
+    };
+}
+/** Record the current declaration as the accepted baseline (`sungen audit --accept-viewpoint`). */
+function acceptViewpointBaseline(projectRoot, unitId, current) {
+    const file = readBaselineFile(projectRoot);
+    file.units[unitId] = {
+        hash: current.hash,
+        ids: current.ids,
+        recordedAt: new Date().toISOString(),
+    };
+    writeBaselineFile(projectRoot, file);
+}
+/** Read the declaration without recording anything (used by `--accept-viewpoint`). */
+function readViewpointDeclaration(viewpointPath) {
+    if (!fs.existsSync(viewpointPath))
+        return null;
+    const ids = (0, parse_1.parseViewpointOverview)(viewpointPath).map((v) => v.id).sort();
+    if (ids.length === 0)
+        return null;
+    return { hash: fingerprint(ids), ids };
+}
+//# sourceMappingURL=viewpoint-baseline.js.map

package/dist/harness/viewpoint-baseline.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"viewpoint-baseline.js","sourceRoot":"","sources":["../../src/harness/viewpoint-baseline.ts"],"names":[],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA+EA,wDAyBC;AAGD,0DAYC;AAGD,4DAKC;AA/HD,+CAAiC;AACjC,uCAAyB;AACzB,2CAA6B;AAC7B,mCAAiD;AAEjD;;;;;;;;;;;;;;;;;;GAkBG;AAEH,+FAA+F;AAClF,QAAA,wBAAwB,GAAG,CAAC,cAAc,EAAE,cAAc,CAAU,CAAC;AAkBlF,SAAS,YAAY,CAAC,WAAmB;IACvC,OAAO,IAAI,CAAC,IAAI,CAAC,WAAW,EAAE,SAAS,EAAE,yBAAyB,CAAC,CAAC;AACtE,CAAC;AAED,SAAS,gBAAgB,CAAC,WAAmB;IAC3C,MAAM,CAAC,GAAG,YAAY,CAAC,WAAW,CAAC,CAAC;IACpC,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,CAAC,CAAC;QAAE,OAAO,EAAE,OAAO,EAAE,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC;IACxD,IAAI,CAAC;QACH,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC,YAAY,CAAC,CAAC,EAAE,OAAO,CAAC,CAAiB,CAAC;QACvE,OAAO,MAAM,IAAI,OAAO,MAAM,KAAK,QAAQ,IAAI,MAAM,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC;IACnG,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,EAAE,OAAO,EAAE,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,CAAG,gDAAgD;IACtF,CAAC;AACH,CAAC;AAED,SAAS,iBAAiB,CAAC,WAAmB,EAAE,IAAkB;IAChE,MAAM,CAAC,GAAG,YAAY,CAAC,WAAW,CAAC,CAAC;IACpC,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;IACnD,EAAE,CAAC,aAAa,CAAC,CAAC,EAAE,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;AACrE,CAAC;AAED;;;GAGG;AACH,SAAS,WAAW,CAAC,GAAa;IAChC,OAAO,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;AACrF,CAAC;AAED;;;;;GAKG;AACH,SAAgB,sBAAsB,CACpC,WAAmB,EACnB,MAAc,EACd,aAAqB;IAErB,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,aAAa,CAAC;QAAE,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,CAAC;IAClF,MAAM,GAAG,GAAG,IAAA,8BAAsB,EAAC,aAAa,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;IAC1E,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC;IAC9B,IAAI,GAAG,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,IAAI,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,CAAC;IAErE,MAAM,IAAI,GAAG,gBAAgB,CAAC,WAAW,CAAC,CAAC;IAC3C,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;IACpC,IAAI,CAAC,QAAQ;QAAE,OAAO,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC;IACnD,IAAI,QAAQ,CAAC,IAAI,KAAK,IAAI;QAAE,OAAO,EAAE,MAAM,EAAE,WAAW,EAAE,IAAI,EAAE,GAAG,EAAE,UAAU,EAAE,QAAQ,CAAC,UAAU,EAAE,CAAC;IAEvG,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,QAAQ,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC;IAC3C,MAAM,GAAG,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC;IACzB,OAAO;QACL,MAAM,EAAE,SAAS;QACjB,IAAI;QACJ,GAAG;QACH,KAAK,EAAE,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;QACxC,OAAO,EAAE,CAAC,QAAQ,CAAC,GAAG,IAAI,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;QACxD,UAAU,EAAE,QAAQ,CAAC,UAAU;KAChC,CAAC;AACJ,CAAC;AAED,mGAAmG;AACnG,SAAgB,uBAAuB,CACrC,WAAmB,EACnB,MAAc,EACd,OAAwC;IAExC,MAAM,IAAI,GAAG,gBAAgB,CAAC,WAAW,CAAC,CAAC;IAC3C,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,GAAG;QACnB,IAAI,EAAE,OAAO,CAAC,IAAI;QAClB,GAAG,EAAE,OAAO,CAAC,GAAG;QAChB,UAAU,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;KACrC,CAAC;IACF,iBAAiB,CAAC,WAAW,EAAE,IAAI,CAAC,CAAC;AACvC,CAAC;AAED,sFAAsF;AACtF,SAAgB,wBAAwB,CAAC,aAAqB;IAC5D,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,aAAa,CAAC;QAAE,OAAO,IAAI,CAAC;IAC/C,MAAM,GAAG,GAAG,IAAA,8BAAsB,EAAC,aAAa,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;IAC1E,IAAI,GAAG,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,IAAI,CAAC;IAClC,OAAO,EAAE,IAAI,EAAE,WAAW,CAAC,GAAG,CAAC,EAAE,GAAG,EAAE,CAAC;AACzC,CAAC"}

package/dist/orchestrator/templates/ai-src/commands/create-test.md

@@ -115,12 +115,16 @@ If the unit is **api-first** (`qa/api/<name>/` or `qa/api/flows/<name>/`), the d
      - **Fill test-viewpoint.md first** — I'll help you identify edge cases, known issues, and design decisions for this screen before generating tests
      - **Continue without it** — generate tests from spec and other sources only
 
+   **A filled `test-viewpoint.md` is an INPUT — never an output.** Do NOT rewrite, re-order, or replace its viewpoint declarations as part of generating tests. It is the yardstick two scored axes measure the suite against (`traceability`, `atomicLedger`): rewrite it to match what you just generated and both read 100% by construction, while any viewpoint you quietly dropped stops being missing from anything. If the declared taxonomy genuinely does not fit the screen, **report the mismatch and ask** — propose the diff (ids added / removed and why) and let the QA decide. `sungen audit` reports `VIEWPOINT-BASELINE-CHANGED` and stops scoring both axes when this file moves, so an unannounced rewrite lowers the score rather than raising it.
+
    **Context discovery (prefer an isolated agent).** Reading all sources here can flood this context. **Claude Code:** spawn the **`sungen-discovery`** sub-agent (Task tool, `subagent_type: sungen-discovery`) to read spec/figma/ui/live in isolation and return a **compact discovery report** (sources, completeness, conflicts, recommended route, key facts); use that report instead of pasting raw sources. **Copilot / no sub-agents:** do the reading inline as below.
 {{/cap}}
 {{^cap parallel-subagents}}
    - If `test-viewpoint.md` exists → read it. If it only contains HTML comments (scaffold template), ask:
      - **1) Fill test-viewpoint.md first** — identify edge cases, known issues, and design decisions before generating tests
      - **2) Continue without it** — generate tests from spec and other sources only
+
+   **A filled `test-viewpoint.md` is an INPUT — never an output.** Do NOT rewrite, re-order, or replace its viewpoint declarations while generating tests. It is the yardstick two scored axes measure the suite against (`traceability`, `atomicLedger`): rewrite it to match what you just generated and both read 100% by construction, while any viewpoint you quietly dropped stops being missing from anything. If the declared taxonomy genuinely does not fit the screen, **report the mismatch and ask** — propose the diff (ids added / removed and why) and let the QA decide. `sungen audit` reports `VIEWPOINT-BASELINE-CHANGED` and stops scoring both axes when this file moves.
 {{/cap}}
 
    **Auto-detect visual source** — do NOT ask the user to pick a source. Instead, check what already exists and use it:

package/dist/orchestrator/templates/ai-src/skills/sungen-tc-generation/SKILL.md

@@ -340,6 +340,7 @@ Security:         [S1 – admin only]
 `sungen audit` enforces these. Generate compliant output up front:
 
 1. **Taxonomy-match** (`VP-TAXONOMY-MISMATCH`, gate-FAIL) — when `test-viewpoint.md` declares its own viewpoint IDs (e.g. `VP0`, `VP1`, … `VP12`, `MS-HP-001`, `MS-EH-001`), **reuse those IDs verbatim as the scenario codes**. Do NOT invent a generic `VP-UI / VP-LOGIC / VP-VAL` scheme — that breaks the coverage matrix. Only fall back to `VP-<CATEGORY>-<NNN>` when the viewpoint file declares no IDs.
+   - **Match the scenarios to the file — never the file to the scenarios.** A filled `test-viewpoint.md` is an input; do not rewrite its declarations to fit what you generated. That is not compliance, it is moving the yardstick: `traceability` + `atomicLedger` then read 100% by construction and a dropped viewpoint stops being missing from anything. Disagree with the taxonomy → propose the diff and ask. `sungen audit` reports `VIEWPOINT-BASELINE-CHANGED` and excludes both axes until a human accepts the change (`sungen audit --screen <name> --accept-viewpoint`).
 2. **Spec-coverage triggers** (`TRIGGER-UNCOVERED`, gate-FAIL) — the Validation-Rules table lists a **trigger** per constraint (e.g. `blur, submit`). Generate one scenario **per (constraint × trigger)** — a `format` rule validating *on blur AND on submit* needs BOTH a blur scenario (`press Tab`) and a submit scenario (`click [Submit]` / `press Enter`). Never collapse the trigger × input matrix to one representative case.
 3. **Claim-Proof** (`CLAIM-UNPROVEN`) — a title claiming `all`/`only`/`every`/`single`/`correct`/`same`/`changes`/`hidden`/`cleared`/`restored`/`independent`/`sanitized`/`announces` MUST have the matching assertion (`see all …`, count, `remember`+compare, `is hidden`, return-and-assert-empty, etc.). If the title promises it, the steps must prove it.
    - **Negative / absence claims** (`does not` / `no` / `never` / `prevents` / `không` / `chưa` — any language; `no-side-effect/no-duplicate`, `negative-claim/absence`): the `Then` must **differ** between the claim holding and not holding. A terminal `see [X] page` that looks identical whether or not the bad thing happened proves nothing. For a side-effect that should NOT repeat (re-submit on back, re-charge, duplicate order, resend OTP), assert the **count is unchanged** (`User see [Records] table with {{one}}` / `row with {{count}}`); if it's not UI-observable, mark `@manual` with a request-count oracle (shape below). This is general — it covers any side-effect, not a fixed verb list.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sun-asterisk/sungen",
-  "version": "3.2.16-beta.12",
+  "version": "3.2.16-beta.13",
   "description": "Deterministic E2E Test Compiler - Gherkin + Selectors → Playwright tests",
   "main": "src/index.ts",
   "types": "src/index.ts",
@@ -12,7 +12,7 @@
     "copy-templates": "mkdir -p dist/generators/test-generator/adapters/playwright/templates/steps && mkdir -p dist/generators/test-generator/templates && mkdir -p dist/orchestrator/templates && mkdir -p dist/dashboard/templates && cp -r src/generators/test-generator/adapters/playwright/templates/*.hbs dist/generators/test-generator/adapters/playwright/templates/ 2>/dev/null || true && cp -r src/generators/test-generator/adapters/playwright/templates/steps dist/generators/test-generator/adapters/playwright/templates/ && mkdir -p dist/generators/test-generator/adapters/appium/templates/steps && cp -r src/generators/test-generator/adapters/appium/templates/*.hbs dist/generators/test-generator/adapters/appium/templates/ 2>/dev/null || true && cp -r src/generators/test-generator/adapters/appium/templates/steps dist/generators/test-generator/adapters/appium/templates/ && cp src/generators/test-generator/templates/*.hbs dist/generators/test-generator/templates/ 2>/dev/null || true && cp -r src/orchestrator/templates/* dist/orchestrator/templates/ && cp src/dashboard/templates/index.html dist/dashboard/templates/index.html && mkdir -p dist/harness/catalog && cp src/harness/catalog/*.yaml dist/harness/catalog/",
     "build:dashboard": "cd ../../dashboard && npm install --silent && npm run build && cd - && cp ../../dashboard/dist/index.html src/dashboard/templates/index.html",
     "dev": "tsx src/cli/index.ts",
-    "test": "tsx tests/golden/run.ts && tsx tests/audit/run.ts && tsx tests/ingest/run.ts && tsx tests/eval/run.ts && tsx tests/exporter/run.ts && tsx tests/exporter/delivery-matrix.run.ts && tsx tests/exporter/delivery-scope-strictness.run.ts && tsx tests/exporter/feature-parser-category.run.ts && tsx tests/exporter/api-detail-sheet.run.ts && tsx tests/exporter/overview-sheet.run.ts && tsx tests/exporter/delivery-divider-tcid.run.ts && tsx tests/exporter/note-expected-actual.run.ts && tsx tests/exporter/test-data-vars.run.ts && tsx tests/exporter/mobile-app-id.run.ts && tsx tests/exporter/api-testcase-cells.run.ts && tsx tests/exporter/delivery-cases-result.run.ts && tsx tests/exporter/precondition-auth-role.run.ts && tsx tests/exporter/report-builder-parity.run.ts && tsx tests/exporter/delivery-preflight-testdata.run.ts && tsx tests/exporter/delivery-preflight-selectors.run.ts && tsx tests/exporter/delivery-prerun-overview.run.ts && tsx tests/exporter/delivery-locale-env-discovery.run.ts && tsx tests/exporter/manual-testcase-testdata.run.ts && tsx tests/dashboard/api-flows-discovery.run.ts && tsx tests/dashboard/status-results-fallback.run.ts && tsx tests/codegen/verb-ref-masking.run.ts && tsx tests/codegen/table-aria-hidden-locators.run.ts && tsx tests/codegen/steps-base-not-compiled.run.ts && tsx tests/codegen/manual-oracle-gate.run.ts && tsx tests/codegen/locator-value-qualified-assertions.run.ts && tsx tests/codegen/table-match-data-column-mapping.run.ts && tsx tests/codegen/param-binding-native-type.run.ts && tsx tests/codegen/table-row-scope-redeclaration.run.ts && tsx tests/codegen/table-row-label-resolution.run.ts && tsx tests/api-runtime/base-path-url-join.run.ts && tsx tests/api-runtime/file-upload-multipart.run.ts && tsx tests/api-runtime/schema-assertion.run.ts && tsx tests/runtime/test-data-cross-ref.run.ts && tsx tests/runtime/test-data-array-projection.run.ts && tsx tests/runtime/env-secret-interpolation.run.ts && tsx tests/capabilities/run.ts && tsx tests/openapi/run.ts && tsx tests/api-field-coverage/run.ts && tsx tests/packaging/run.ts && tsx tests/generate-hint/run.ts && tsx tests/template-assertion/run.ts && tsx tests/journey/run.ts && tsx tests/harness/serial-cascade.run.ts && tsx tests/harness/score-model.run.ts && tsx tests/harness/spec-trace-crlf.run.ts && tsx tests/ai-skills/cross-assistant-orphan.run.ts && tsx tests/init/codex-mcp-config.run.ts && tsx src/orchestrator/ai-skills/golden-skills.test.ts && tsx tests/db-runtime/sql-placeholder-rewrite.run.ts && tsx tests/db-runtime/fallback-flow-codified.run.ts && tsx tests/db-runtime/cosmos-engine.run.ts && tsx tests/db-runtime/mongodb-engine.run.ts && tsx tests/db-runtime/dynamodb-engine.run.ts && tsx tests/db-runtime/mysql-integration.opt-in.run.ts && tsx tests/db-runtime/mongodb-integration.opt-in.run.ts && tsx tests/db-runtime/dynamodb-integration.opt-in.run.ts",
+    "test": "tsx tests/golden/run.ts && tsx tests/audit/run.ts && tsx tests/ingest/run.ts && tsx tests/eval/run.ts && tsx tests/exporter/run.ts && tsx tests/exporter/delivery-matrix.run.ts && tsx tests/exporter/delivery-scope-strictness.run.ts && tsx tests/exporter/feature-parser-category.run.ts && tsx tests/exporter/api-detail-sheet.run.ts && tsx tests/exporter/overview-sheet.run.ts && tsx tests/exporter/delivery-divider-tcid.run.ts && tsx tests/exporter/note-expected-actual.run.ts && tsx tests/exporter/test-data-vars.run.ts && tsx tests/exporter/mobile-app-id.run.ts && tsx tests/exporter/api-testcase-cells.run.ts && tsx tests/exporter/delivery-cases-result.run.ts && tsx tests/exporter/precondition-auth-role.run.ts && tsx tests/exporter/report-builder-parity.run.ts && tsx tests/exporter/delivery-preflight-testdata.run.ts && tsx tests/exporter/delivery-preflight-selectors.run.ts && tsx tests/exporter/delivery-prerun-overview.run.ts && tsx tests/exporter/delivery-locale-env-discovery.run.ts && tsx tests/exporter/manual-testcase-testdata.run.ts && tsx tests/dashboard/api-flows-discovery.run.ts && tsx tests/dashboard/status-results-fallback.run.ts && tsx tests/codegen/verb-ref-masking.run.ts && tsx tests/codegen/table-aria-hidden-locators.run.ts && tsx tests/codegen/steps-base-not-compiled.run.ts && tsx tests/codegen/manual-oracle-gate.run.ts && tsx tests/codegen/locator-value-qualified-assertions.run.ts && tsx tests/codegen/table-match-data-column-mapping.run.ts && tsx tests/codegen/param-binding-native-type.run.ts && tsx tests/codegen/table-row-scope-redeclaration.run.ts && tsx tests/codegen/table-row-label-resolution.run.ts && tsx tests/api-runtime/base-path-url-join.run.ts && tsx tests/api-runtime/file-upload-multipart.run.ts && tsx tests/api-runtime/schema-assertion.run.ts && tsx tests/runtime/test-data-cross-ref.run.ts && tsx tests/runtime/test-data-array-projection.run.ts && tsx tests/runtime/env-secret-interpolation.run.ts && tsx tests/capabilities/run.ts && tsx tests/openapi/run.ts && tsx tests/api-field-coverage/run.ts && tsx tests/packaging/run.ts && tsx tests/generate-hint/run.ts && tsx tests/template-assertion/run.ts && tsx tests/journey/run.ts && tsx tests/harness/serial-cascade.run.ts && tsx tests/harness/score-model.run.ts && tsx tests/harness/spec-trace-crlf.run.ts && tsx tests/harness/viewpoint-baseline.run.ts && tsx tests/ai-skills/cross-assistant-orphan.run.ts && tsx tests/init/codex-mcp-config.run.ts && tsx src/orchestrator/ai-skills/golden-skills.test.ts && tsx tests/db-runtime/sql-placeholder-rewrite.run.ts && tsx tests/db-runtime/fallback-flow-codified.run.ts && tsx tests/db-runtime/cosmos-engine.run.ts && tsx tests/db-runtime/mongodb-engine.run.ts && tsx tests/db-runtime/dynamodb-engine.run.ts && tsx tests/db-runtime/mysql-integration.opt-in.run.ts && tsx tests/db-runtime/mongodb-integration.opt-in.run.ts && tsx tests/db-runtime/dynamodb-integration.opt-in.run.ts",
     "test:update": "tsx tests/golden/run.ts --update && tsx tests/audit/run.ts --update && tsx tests/ingest/run.ts --update",
     "prepublishOnly": "npm run build:dashboard && npm run build"
   },
@@ -39,8 +39,8 @@
     "@babel/types": "^7.28.5",
     "@cucumber/gherkin": "^37.0.0",
     "@cucumber/messages": "^31.0.0",
-    "@sungen/driver-data-factory": "3.2.16-beta.12",
-    "@sungen/driver-ui": "3.2.16-beta.12",
+    "@sungen/driver-data-factory": "3.2.16-beta.13",
+    "@sungen/driver-ui": "3.2.16-beta.13",
     "chalk": "^5.6.2",
     "commander": "^14.0.2",
     "dotenv": "^17.2.3",

package/src/cli/commands/audit.ts

@@ -1,8 +1,9 @@
 import { Command } from 'commander';
 import * as path from 'path';
 import * as fs from 'fs';
-import { runAudit, AuditReport } from '../../harness/audit';
+import { runAudit, AuditReport, catalogIdFromScreenDir } from '../../harness/audit';
 import { reportSlug } from '../../harness/unit-paths';
+import { acceptViewpointBaseline } from '../../harness/viewpoint-baseline';
 
 function findScreenDir(name: string): string | null {
   // `name` may be a bare screen/flow/area, or an api unit id (`api/<area>`, `api/flows/<flow>`).
@@ -100,6 +101,7 @@ export function registerAuditCommand(program: Command): void {
     .option('--api <name>', 'API-first area or api flow to audit (e.g. orders, flows/signup)')
     .option('--area <name>', 'Alias of --api — an API-first area (qa/api/<name>)')
     .option('--json', 'Output the raw JSON report')
+    .option('--accept-viewpoint', 'Accept the current test-viewpoint.md as the baseline the suite is measured against')
     .action((options) => {
       try {
         const name = options.screen || options.api || options.area;
@@ -109,6 +111,21 @@ export function registerAuditCommand(program: Command): void {
 
         const report = runAudit(dir, name);
 
+        // The baseline is only ever written HERE, never by the measurement itself:
+        // a first sighting has nothing to compare against, and any later change is
+        // a human decision (`--accept-viewpoint`) — see harness/viewpoint-baseline.ts.
+        const unitId = catalogIdFromScreenDir(dir);
+        const vb = report.viewpointBaseline;
+        if (vb.status === 'new' || (options.acceptViewpoint && vb.status === 'changed')) {
+          acceptViewpointBaseline(process.cwd(), unitId, { hash: vb.hash, ids: vb.ids });
+          if (options.acceptViewpoint && !options.json) {
+            console.log(`\n  ✓ test-viewpoint.md accepted as the baseline for ${name} (${vb.ids.length} viewpoint id(s)).`);
+            console.log('    Re-run the audit to score atomicLedger + traceability against it.\n');
+          }
+        } else if (options.acceptViewpoint && !options.json) {
+          console.log(`\n  ○ nothing to accept for ${name} — the viewpoint baseline is ${vb.status}.\n`);
+        }
+
         // Persist report under .sungen/reports/ (flat slug: api flow `flows/x` → `flows-x-audit.json`).
         const outDir = path.join(process.cwd(), '.sungen', 'reports');
         fs.mkdirSync(outDir, { recursive: true });

package/src/exporters/matrix/build.ts

@@ -407,6 +407,22 @@ function commonPrefix(lists: string[][]): number {
   return n;
 }
 
+/**
+ * What the parent row's Action says when the variants share no opening step.
+ *
+ * The parent is the row a reviewer (and the customer) reads AS the test case, and
+ * with nothing to hoist its Action cell came out empty — 12 of 38 items in a real
+ * delivery. The steps are not missing, they are per-variant; say that, and name the
+ * axis they vary along so the cell carries meaning on its own.
+ */
+export function varyingTriggerNote(dimensions: string | undefined, count: number): string {
+  // `dimensions` reads "matching rule — title only · content only · …"; the part
+  // before the dash is the axis name.
+  const axis = dimensions?.split('—')[0].trim();
+  const where = axis && axis.length > 0 && axis.length <= 60 ? ` by ${axis}` : '';
+  return `The action differs${where} — run each of the ${count} variants below with its own steps.`;
+}
+
 /**
  * Expand one map variant ref to concrete variant refs:
  * a bare VP-id whose scenario has a dataset means ALL its rows.
@@ -471,7 +487,10 @@ export function buildMatrix(inputs: BuildInputs): MatrixModel {
         [v.ref, v.precondition.filter((line) => !commonPre.includes(line))])),
       triggerDeltas: Object.fromEntries(groupVariants.map((v) =>
         [v.ref, v.trigger.slice(prefixLen)])),
-      trigger: (first?.trigger ?? []).slice(0, prefixLen),
+      // Nothing shared to hoist → say what varies rather than leaving the cell blank.
+      trigger: prefixLen > 0 || groupVariants.length === 0
+        ? (first?.trigger ?? []).slice(0, prefixLen)
+        : [varyingTriggerNote(g.dimensions, groupVariants.length)],
       variants: groupVariants,
       result,
       resultCounts: counts,

package/src/harness/audit.ts

@@ -9,6 +9,7 @@
 import * as path from 'path';
 import * as fs from 'fs';
 import { loadScenarios, parseViewpointOverview, ScenarioInfo, ViewpointEntry } from './parse';
+import { checkViewpointBaseline, ViewpointBaseline } from './viewpoint-baseline';
 import { featureBasename } from './unit-paths';
 import {
   loadCatalog, viewpointGate, assertionDepth, dataThemesFor, depthThresholdFor, coverageBalance, duplicateClusters, traceability, claimProof, taxonomyLint,
@@ -46,6 +47,7 @@ export interface AuditReport {
   flowDepth: FlowDepthResult;   // H3 — stateful-flow regression depth (count / teardown / multi-source)
   oracle: OracleStrengthResult; // H4 — facet-oracle strength (weak name-substring vs title/detail/API/DB)
   ledger: LedgerResult;         // atomic viewpoint-item coverage (per-bullet status)
+  viewpointBaseline: ViewpointBaseline; // is the yardstick still the accepted one? (#557)
   calibration: {                // #8 — multi-axis score so a high overall can't hide a weak axis
     axes: Record<string, number>;
     weakest: { axis: string; value: number };
@@ -73,7 +75,7 @@ export interface AuditReport {
 }
 
 /** The catalog-resolution id for a unit dir (relative to qa/): screen · flows/<flow> · api/<area> · api/flows/<flow>. */
-function catalogIdFromScreenDir(screenDir: string): string {
+export function catalogIdFromScreenDir(screenDir: string): string {
   const parts = screenDir.split(path.sep);
   const qa = parts.lastIndexOf('qa');
   if (qa >= 0) {
@@ -154,6 +156,9 @@ export function runAudit(screenDir: string, screenName: string): AuditReport {
   const capGate = scoringCap?.gateProvider as
     ((i: { scenarios: ScenarioInfo[]; viewpoints: ViewpointEntry[]; catalog: Catalog; focus: typeof intent.focus; cwd: string; screenName: string; threshold: number; businessCriticalMethods?: string[] }) => { gate: GateResult; depth: DepthResult }) | undefined;
   const provided = capGate?.({ scenarios, viewpoints, catalog, focus: intent.focus, cwd: projectRootFromScreenDir(screenDir), screenName: catalogScreenName, threshold: depthThresholdFor(intent.focus), businessCriticalMethods: intent.businessCriticalMethods });
+  // Is the declaration this suite is measured against still the one a human accepted?
+  const viewpointBaseline = checkViewpointBaseline(
+    projectRootFromScreenDir(screenDir), catalogScreenName, viewpointPath);
   const viewpointText = fs.existsSync(viewpointPath) ? readTextFile(viewpointPath) : '';
   const gate = provided?.gate ?? viewpointGate(scenarios, viewpoints, catalog, platform === 'mobile', viewpointText);
   const depth = provided?.depth ?? assertionDepth(scenarios, dataThemesFor(catalog, gate.pageType), intent.focus);
@@ -234,13 +239,19 @@ export function runAudit(screenDir: string, screenName: string): AuditReport {
   //     claimProof, specFR, atomicLedger — now carry weight. A suite where two
   //     thirds of the scenarios do not prove their own title used to score 7.9.
   const specRatio = spec.frTotal ? spec.frCovered / spec.frTotal : 1;
+  //  3. Evidence has to be INDEPENDENT of what it judges. `atomicLedger` and
+  //     `traceability` both measure the suite against test-viewpoint.md, so a run
+  //     that rewrites that file scores them 100% by construction — which is exactly
+  //     what a create-test run did, silently dropping the performance viewpoint on
+  //     the way. While the declaration is unconfirmed, neither axis is evidence.
+  const viewpointMoved = viewpointBaseline.status === 'changed';
   const axisDefs: Array<{ key: string; value: number; weight: number; applicable: boolean; critical: boolean }> = [
     { key: 'coverage', value: coverage, weight: 0.22, applicable: !!gate.pageType && gate.themesTotal > 0, critical: true },
     { key: 'specFR', value: specRatio, weight: 0.15, applicable: spec.hasSpec && spec.frTotal > 0, critical: true },
-    { key: 'atomicLedger', value: ledger.ratio, weight: 0.13, applicable: ledger.hasViewpoint && ledger.total > 0, critical: true },
+    { key: 'atomicLedger', value: ledger.ratio, weight: 0.13, applicable: ledger.hasViewpoint && ledger.total > 0 && !viewpointMoved, critical: true },
     { key: 'businessDepth', value: businessDepth, weight: 0.20, applicable: true, critical: true },
     { key: 'claimProof', value: claim.ratio, weight: 0.15, applicable: claim.withClaims > 0, critical: true },
-    { key: 'traceability', value: traceScore, weight: 0.09, applicable: viewpoints.length > 0, critical: false },
+    { key: 'traceability', value: traceScore, weight: 0.09, applicable: viewpoints.length > 0 && !viewpointMoved, critical: false },
     { key: 'balance', value: balanceScore, weight: 0.06, applicable: true, critical: false },
   ];
   const scored = axisDefs.filter((a) => a.applicable);
@@ -278,6 +289,13 @@ export function runAudit(screenDir: string, screenName: string): AuditReport {
       : `this unit supplies no evidence for [${missingEvidence.join(', ')}], and a top mark has to rest on complete evidence`;
     findings.push(`SCORE-CAPPED: overall held at ${cap.toFixed(1)} because ${why}.`);
   }
+  if (viewpointBaseline.status === 'changed') {
+    const moved = [
+      viewpointBaseline.removed?.length ? `removed [${viewpointBaseline.removed.join(', ')}]` : '',
+      viewpointBaseline.added?.length ? `added [${viewpointBaseline.added.join(', ')}]` : '',
+    ].filter(Boolean).join(', ');
+    findings.push(`VIEWPOINT-BASELINE-CHANGED: test-viewpoint.md no longer declares what it did when this unit was last accepted — ${moved || 'the declared ids were reordered or replaced'}. atomicLedger + traceability measure the suite AGAINST this file, so they are excluded from the score until the change is confirmed: a generator that rewrites the declaration scores both 100% by construction, and a viewpoint dropped from the file stops being missing from anything. Review the diff (a removed id means that coverage is now unclaimed), then run \`sungen audit --screen ${screenName} --accept-viewpoint\`.`);
+  }
   if (gate.pageTypeSource === 'undetermined') {
     findings.push(`PAGE-TYPE-UNDETERMINED: no page type fits this screen with enough confidence (best ${gate.pageTypeEvidence?.hits ?? 0} keyword hit(s)), so NO critical themes were demanded and the coverage axis is excluded from the score — declare it in test-viewpoint.md (\`page-type: <id>\`) to have the theme checklist applied.`);
   }
@@ -465,7 +483,7 @@ export function runAudit(screenDir: string, screenName: string): AuditReport {
     screen: screenName,
     scenarioCount: scenarios.length,
     gate, depth, claim, taxonomy, balance, duplicates, trace, spec,
-    taxonomyMismatch, downstream, manualOracle: manualOracleResult, automatableManual: autoManual, flowDepth, oracle, ledger, calibration,
+    taxonomyMismatch, downstream, manualOracle: manualOracleResult, automatableManual: autoManual, flowDepth, oracle, ledger, viewpointBaseline, calibration,
     score: {
       overall: Math.round(overall * 10) / 10,
       coverage: Math.round(coverage * 100) / 100,

package/src/harness/capability-plan.ts

@@ -114,6 +114,11 @@ export function inferFromText(reason: string): string | undefined {
   return undefined;
 }
 
+/** Every code the text supports, not just the first rule in declaration order. */
+export function inferAllFromText(reason: string): string[] {
+  return INFER.filter((r) => r.re.test(reason)).map((r) => r.code);
+}
+
 export interface ReasonMismatch { scenario: string; explicit: string; inferred: string }
 
 /**
@@ -150,8 +155,15 @@ export function manualReasonMismatches(featurePath: string): ReasonMismatch[] {
       else if (l === '') continue;
       else break;
     }
-    const inferred = inferFromText(parts.join(' '));
-    if (inferred && inferred !== explicit) out.push({ scenario: m[1].trim(), explicit, inferred });
+    // The block scanned here includes the tester PROCEDURE, so several codes can
+    // match at once — "not worth automating" (M8, the author's actual reason) and
+    // "the service stays responsive" (M6, a word about latency, not layout). The
+    // first rule in declaration order used to win and the correct tag was reported
+    // as a mismatch. A tag is only wrong when the text supports some OTHER code and
+    // NOT the declared one.
+    const codes = inferAllFromText(parts.join(' '));
+    const inferred = codes.find((c) => c !== explicit);
+    if (inferred && !codes.includes(explicit)) out.push({ scenario: m[1].trim(), explicit, inferred });
   }
   return out;
 }

package/src/harness/viewpoint-baseline.ts

@@ -0,0 +1,128 @@
+import * as crypto from 'crypto';
+import * as fs from 'fs';
+import * as path from 'path';
+import { parseViewpointOverview } from './parse';
+
+/**
+ * Keep the yardstick from moving with the thing it measures.
+ *
+ * `test-viewpoint.md` is the project's declaration of WHAT must be tested, and two
+ * scored axes measure the suite against it: `traceability` (do scenarios use the
+ * declared viewpoint ids?) and `atomicLedger` (is every declared item covered?).
+ * Both are only evidence while the declaration is INDEPENDENT of the suite.
+ *
+ * A `/sungen:create-test` run rewrote a filled test-viewpoint.md, replacing the
+ * declared categories with exactly the ones it had just generated. Both axes read
+ * 100% by construction — and the performance viewpoint it dropped on the way
+ * (4 perf scenarios → 2) went unreported, because VP-PERF was no longer declared
+ * for anything to be missing from.
+ *
+ * So: remember the declaration, and when it changes, say so and stop counting the
+ * two axes that depend on it until a human accepts the new baseline. Same rule the
+ * score model already applies to an undetermined page type — absent independent
+ * evidence scores nothing, in either direction.
+ */
+
+/** Axes whose evidence is `test-viewpoint.md` — excluded while the baseline is unconfirmed. */
+export const VIEWPOINT_DEPENDENT_AXES = ['atomicLedger', 'traceability'] as const;
+
+export interface ViewpointBaseline {
+  status: 'absent' | 'new' | 'unchanged' | 'changed';
+  /** sha1 of the normalized file — '' when the unit has no test-viewpoint.md. */
+  hash: string;
+  ids: string[];
+  /** Only on `changed`: how the declaration moved. */
+  added?: string[];
+  removed?: string[];
+  recordedAt?: string;
+}
+
+interface BaselineFile {
+  version: 1;
+  units: Record<string, { hash: string; ids: string[]; recordedAt: string }>;
+}
+
+function baselinePath(projectRoot: string): string {
+  return path.join(projectRoot, '.sungen', 'viewpoint-baseline.json');
+}
+
+function readBaselineFile(projectRoot: string): BaselineFile {
+  const p = baselinePath(projectRoot);
+  if (!fs.existsSync(p)) return { version: 1, units: {} };
+  try {
+    const parsed = JSON.parse(fs.readFileSync(p, 'utf-8')) as BaselineFile;
+    return parsed && typeof parsed === 'object' && parsed.units ? parsed : { version: 1, units: {} };
+  } catch {
+    return { version: 1, units: {} };   // unreadable → treat as unrecorded, never throw
+  }
+}
+
+function writeBaselineFile(projectRoot: string, data: BaselineFile): void {
+  const p = baselinePath(projectRoot);
+  fs.mkdirSync(path.dirname(p), { recursive: true });
+  fs.writeFileSync(p, `${JSON.stringify(data, null, 2)}\n`, 'utf-8');
+}
+
+/**
+ * Hash the DECLARATION, not the prose: an author reflowing a sentence must not read
+ * as the taxonomy moving. Only the declared viewpoint ids are fingerprinted.
+ */
+function fingerprint(ids: string[]): string {
+  return crypto.createHash('sha1').update(ids.join('\n')).digest('hex').slice(0, 12);
+}
+
+/**
+ * Compare the unit's current test-viewpoint declaration to the recorded baseline.
+ * Read-only — recording is the caller's decision, so an audit stays a measurement
+ * and `--accept-viewpoint` stays the explicit human act. A first sighting reports
+ * `new` (nothing to compare against yet).
+ */
+export function checkViewpointBaseline(
+  projectRoot: string,
+  unitId: string,
+  viewpointPath: string,
+): ViewpointBaseline {
+  if (!fs.existsSync(viewpointPath)) return { status: 'absent', hash: '', ids: [] };
+  const ids = parseViewpointOverview(viewpointPath).map((v) => v.id).sort();
+  const hash = fingerprint(ids);
+  if (ids.length === 0) return { status: 'absent', hash: '', ids: [] };
+
+  const file = readBaselineFile(projectRoot);
+  const recorded = file.units[unitId];
+  if (!recorded) return { status: 'new', hash, ids };
+  if (recorded.hash === hash) return { status: 'unchanged', hash, ids, recordedAt: recorded.recordedAt };
+
+  const before = new Set(recorded.ids ?? []);
+  const now = new Set(ids);
+  return {
+    status: 'changed',
+    hash,
+    ids,
+    added: ids.filter((i) => !before.has(i)),
+    removed: (recorded.ids ?? []).filter((i) => !now.has(i)),
+    recordedAt: recorded.recordedAt,
+  };
+}
+
+/** Record the current declaration as the accepted baseline (`sungen audit --accept-viewpoint`). */
+export function acceptViewpointBaseline(
+  projectRoot: string,
+  unitId: string,
+  current: { hash: string; ids: string[] },
+): void {
+  const file = readBaselineFile(projectRoot);
+  file.units[unitId] = {
+    hash: current.hash,
+    ids: current.ids,
+    recordedAt: new Date().toISOString(),
+  };
+  writeBaselineFile(projectRoot, file);
+}
+
+/** Read the declaration without recording anything (used by `--accept-viewpoint`). */
+export function readViewpointDeclaration(viewpointPath: string): { hash: string; ids: string[] } | null {
+  if (!fs.existsSync(viewpointPath)) return null;
+  const ids = parseViewpointOverview(viewpointPath).map((v) => v.id).sort();
+  if (ids.length === 0) return null;
+  return { hash: fingerprint(ids), ids };
+}

package/src/orchestrator/templates/ai-src/commands/create-test.md

@@ -115,12 +115,16 @@ If the unit is **api-first** (`qa/api/<name>/` or `qa/api/flows/<name>/`), the d
      - **Fill test-viewpoint.md first** — I'll help you identify edge cases, known issues, and design decisions for this screen before generating tests
      - **Continue without it** — generate tests from spec and other sources only
 
+   **A filled `test-viewpoint.md` is an INPUT — never an output.** Do NOT rewrite, re-order, or replace its viewpoint declarations as part of generating tests. It is the yardstick two scored axes measure the suite against (`traceability`, `atomicLedger`): rewrite it to match what you just generated and both read 100% by construction, while any viewpoint you quietly dropped stops being missing from anything. If the declared taxonomy genuinely does not fit the screen, **report the mismatch and ask** — propose the diff (ids added / removed and why) and let the QA decide. `sungen audit` reports `VIEWPOINT-BASELINE-CHANGED` and stops scoring both axes when this file moves, so an unannounced rewrite lowers the score rather than raising it.
+
    **Context discovery (prefer an isolated agent).** Reading all sources here can flood this context. **Claude Code:** spawn the **`sungen-discovery`** sub-agent (Task tool, `subagent_type: sungen-discovery`) to read spec/figma/ui/live in isolation and return a **compact discovery report** (sources, completeness, conflicts, recommended route, key facts); use that report instead of pasting raw sources. **Copilot / no sub-agents:** do the reading inline as below.
 {{/cap}}
 {{^cap parallel-subagents}}
    - If `test-viewpoint.md` exists → read it. If it only contains HTML comments (scaffold template), ask:
      - **1) Fill test-viewpoint.md first** — identify edge cases, known issues, and design decisions before generating tests
      - **2) Continue without it** — generate tests from spec and other sources only
+
+   **A filled `test-viewpoint.md` is an INPUT — never an output.** Do NOT rewrite, re-order, or replace its viewpoint declarations while generating tests. It is the yardstick two scored axes measure the suite against (`traceability`, `atomicLedger`): rewrite it to match what you just generated and both read 100% by construction, while any viewpoint you quietly dropped stops being missing from anything. If the declared taxonomy genuinely does not fit the screen, **report the mismatch and ask** — propose the diff (ids added / removed and why) and let the QA decide. `sungen audit` reports `VIEWPOINT-BASELINE-CHANGED` and stops scoring both axes when this file moves.
 {{/cap}}
 
    **Auto-detect visual source** — do NOT ask the user to pick a source. Instead, check what already exists and use it:

package/src/orchestrator/templates/ai-src/skills/sungen-tc-generation/SKILL.md

@@ -340,6 +340,7 @@ Security:         [S1 – admin only]
 `sungen audit` enforces these. Generate compliant output up front:
 
 1. **Taxonomy-match** (`VP-TAXONOMY-MISMATCH`, gate-FAIL) — when `test-viewpoint.md` declares its own viewpoint IDs (e.g. `VP0`, `VP1`, … `VP12`, `MS-HP-001`, `MS-EH-001`), **reuse those IDs verbatim as the scenario codes**. Do NOT invent a generic `VP-UI / VP-LOGIC / VP-VAL` scheme — that breaks the coverage matrix. Only fall back to `VP-<CATEGORY>-<NNN>` when the viewpoint file declares no IDs.
+   - **Match the scenarios to the file — never the file to the scenarios.** A filled `test-viewpoint.md` is an input; do not rewrite its declarations to fit what you generated. That is not compliance, it is moving the yardstick: `traceability` + `atomicLedger` then read 100% by construction and a dropped viewpoint stops being missing from anything. Disagree with the taxonomy → propose the diff and ask. `sungen audit` reports `VIEWPOINT-BASELINE-CHANGED` and excludes both axes until a human accepts the change (`sungen audit --screen <name> --accept-viewpoint`).
 2. **Spec-coverage triggers** (`TRIGGER-UNCOVERED`, gate-FAIL) — the Validation-Rules table lists a **trigger** per constraint (e.g. `blur, submit`). Generate one scenario **per (constraint × trigger)** — a `format` rule validating *on blur AND on submit* needs BOTH a blur scenario (`press Tab`) and a submit scenario (`click [Submit]` / `press Enter`). Never collapse the trigger × input matrix to one representative case.
 3. **Claim-Proof** (`CLAIM-UNPROVEN`) — a title claiming `all`/`only`/`every`/`single`/`correct`/`same`/`changes`/`hidden`/`cleared`/`restored`/`independent`/`sanitized`/`announces` MUST have the matching assertion (`see all …`, count, `remember`+compare, `is hidden`, return-and-assert-empty, etc.). If the title promises it, the steps must prove it.
    - **Negative / absence claims** (`does not` / `no` / `never` / `prevents` / `không` / `chưa` — any language; `no-side-effect/no-duplicate`, `negative-claim/absence`): the `Then` must **differ** between the claim holding and not holding. A terminal `see [X] page` that looks identical whether or not the bad thing happened proves nothing. For a side-effect that should NOT repeat (re-submit on back, re-charge, duplicate order, resend OTP), assert the **count is unchanged** (`User see [Records] table with {{one}}` / `row with {{count}}`); if it's not UI-observable, mark `@manual` with a request-count oracle (shape below). This is general — it covers any side-effect, not a fixed verb list.