universal-dev-standards 5.15.1 → 5.16.0

package/bundled/skills/commands/brainstorm.md

@@ -6,9 +6,9 @@ argument-hint: "[problem or feature idea | 問題或功能構想]"
 
 # Brainstorm Assistant | 腦力激盪助手
 
-Structured ideation before specification writing. Transform vague ideas into actionable feature proposals through guided brainstorming.
+Structured ideation before specification writing. Transform vague ideas into actionable feature proposals through a persona-ensemble brainstorm.
 
-在撰寫規格前進行結構化發想。透過引導式腦力激盪，將模糊構想轉化為可執行的功能提案。
+在撰寫規格前進行結構化發想。透過 persona 集成式腦力激盪，將模糊構想轉化為可執行的功能提案。
 
 ## Usage | 用法
 
@@ -19,33 +19,37 @@ Structured ideation before specification writing. Transform vague ideas into act
 ## Workflow | 工作流程
 
 ```
-FRAME ──► DIVERGE ──► CONVERGE ──► OUTPUT
-定義問題     發散思考       收斂評估       輸出提案
+PRE-FLIGHT ──► FRAME ──► DIVERGE ──────────► CONVERGE ─────────► OUTPUT
+  防錨定        定義問題    persona 集成+透鏡     多評審面板+硬角色反駁   輸出提案
 ```
 
-| Phase | Goal | Key Techniques | 目標 |
-|-------|------|----------------|------|
+| Phase | Goal | Key Techniques (v3) | 目標 |
+|-------|------|---------------------|------|
+| **PRE-FLIGHT** | Prevent AI anchoring | User writes 3 ideas first; no analogy seed | 防止錨定 |
 | **FRAME** | Define problem clearly | 5 Whys, HMW, Stakeholder Map | 清楚定義問題 |
-| **DIVERGE** | Generate many ideas | HMW, SCAMPER, Six Thinking Hats | 產生大量想法 |
-| **CONVERGE** | Evaluate and rank | Evaluation Matrix, Dot Voting | 評估與排序 |
+| **DIVERGE** | Force viewpoint diversity | Persona ensemble + diversity lenses | 逼出視角多樣性 |
+| **CONVERGE** | Bias-checked selection | Multi-critic panel + Devil's Advocate/Steelman | 降偏誤選擇 |
 | **OUTPUT** | Actionable report | Brainstorm Report template | 可執行的報告 |
 
 ## Techniques | 技法
 
 | Technique | Best For | 適用場景 |
 |-----------|----------|----------|
-| **5 Whys** | Finding root causes | 找到根本原因 |
-| **HMW** | Reframing problems as opportunities | 將問題重構為機會 |
-| **SCAMPER** | Improving existing features | 改善現有功能 |
-| **Six Thinking Hats** | Multi-perspective analysis | 多角度分析 |
-| **Dot Voting** | Quick prioritization | 快速排序 |
+| **Persona ensemble** | Forced viewpoint diversity (v3 core) | 強制視角多樣性（v3 核心） |
+| **Diversity lenses** | Analogical / reversal / morphological | 突破顯而易見區 |
+| **Multi-critic panel** | Bias-reduced scoring (v3 core) | 降偏誤評分（v3 核心） |
+| **Devil's Advocate + Steelman** | Hard-role rebuttal | 硬角色反駁 |
+| **5 Whys / HMW / SCAMPER / Six Hats** | Classic framing & divergence | 經典框定與發散 |
 
 ## Examples | 範例
 
 ```bash
-/brainstorm                           # Start interactive session
-/brainstorm "user retention"          # Brainstorm around a topic
-/brainstorm --technique scamper       # Use a specific technique
+/brainstorm                                          # Start interactive session
+/brainstorm "user retention"                         # Brainstorm around a topic
+/brainstorm --enhanced "user retention"              # Parallel persona ensemble (if host supports it)
+/brainstorm --personas "designer,economist,skeptic"  # Custom personas
+/brainstorm --lens analogical "onboarding"           # Force the analogical lens
+/brainstorm --quick "reduce checkout friction"       # Fast 3-idea mode
 ```
 
 ## Output Format | 輸出格式
@@ -57,9 +61,10 @@ FRAME ──► DIVERGE ──► CONVERGE ──► OUTPUT
 [Refined problem from FRAME phase]
 
 ## Top 3 Recommendations
-1. **[Idea]** — Score: X.X — [Why recommended]
-2. **[Idea]** — Score: X.X — [Why recommended]
-3. **[Idea]** — Score: X.X — [Why recommended]
+1. **[Idea]** — Agg. X.X ✓ Passed rebuttal — Persona: [..] — [Why recommended]
+
+## Diversity Note
+[How many distinct personas/lenses the surviving ideas span]
 
 ## Next Steps
 - [ ] Proceed to `/requirement` with top idea
@@ -83,26 +88,36 @@ After brainstorming, the typical workflow is:
 
 | Input | AI Action |
 |-------|-----------|
-| `/brainstorm` | 詢問要探討的主題或問題，進入 FRAME 階段 |
-| `/brainstorm "topic"` | 以指定主題開始，進入 FRAME 階段 |
-| `/brainstorm --technique <name>` | 使用指定技法開始 DIVERGE 階段 |
+| `/brainstorm` | 啟動 PRE-FLIGHT，請使用者先寫 問題＋3 想法＋反向排除 |
+| `/brainstorm "topic"` | 以指定主題啟動 PRE-FLIGHT，再進入 FRAME |
+| `/brainstorm --personas "a,b,c"` | 以自訂 persona 組進入 DIVERGE |
+| `/brainstorm --lens <name>` | 以指定多樣性透鏡為主進入 DIVERGE |
+| `/brainstorm --enhanced` | 若宿主支援子代理則平行跑 persona/評審，否則靜默退回 baseline |
 
 ### Interaction Script | 互動腳本
 
+#### PRE-FLIGHT Phase
+1. 請使用者先寫 問題（一句）＋3 個初始想法＋最不想要的解法類型
+2. 拒絕「像 X 但給 Y」種子，改寫為底層問題
+
+🛑 **STOP**: 收到三項輸入前不生成任何想法（`--skip-preflight` 例外，顯示錨定警告）
+
 #### FRAME Phase
-1. 釐清問題陳述（使用 5 Whys 或 HMW）
-2. 確認問題定義
+1. 釐清問題陳述（5 Whys 找根因）
+2. 重構為 HMW 問題；識別利害關係人
 
 🛑 **STOP**: 問題定義後等待使用者確認
 
 #### DIVERGE Phase
-1. 使用選定技法生成想法
-2. 鼓勵數量優先於品質
-3. 列出所有想法
+1. 逐一以預設 persona（領域專家／懷疑者／跨域類比者／成本約束者／使用者代言）思維鏈生成想法
+2. **分支隔離**：生成各 persona 時不互相預覽，全部產完才一起呈現
+3. 至少套用一個多樣性透鏡（類比／假設反轉／形態矩陣）
+4. 以多樣性（覆蓋的 persona/透鏡數）而非數量為繼續門檻
 
 #### CONVERGE Phase
-1. 以評估矩陣（可行性、影響力、成本）評分
-2. 排序前 3 名
+1. 以 3 個獨立評審（工程可行性／使用者影響／策略一致）各自評分後聚合
+2. 對前 3 名跑硬角色 Devil's Advocate（論證會失敗）+ Steelman；使用者以 (a)修改/(b)反駁/(c)移除 回應
+3. 排序並標記通過反駁的想法
 
 🛑 **STOP**: 展示 Brainstorm Report 後等待使用者決定下一步
 
@@ -110,7 +125,9 @@ After brainstorming, the typical workflow is:
 
 | Stop Point | 等待內容 |
 |-----------|---------|
+| PRE-FLIGHT 提交前 | 收到使用者三項輸入 |
 | 問題定義後 | 確認問題正確 |
+| 反駁輪每個反對理由 | 使用者 (a)/(b)/(c) 回應 |
 | 報告展示後 | 決定進入 `/requirement` 或 `/sdd` |
 
 ### Error Handling | 錯誤處理
@@ -118,10 +135,14 @@ After brainstorming, the typical workflow is:
 | Error Condition | AI Action |
 |-----------------|-----------|
 | 主題太廣泛 | 引導縮小範圍 |
-| 指定的技法不存在 | 列出可用技法供選擇 |
+| 使用「像 X 但給 Y」種子 | 改寫為底層問題再繼續（反種子 guardrail） |
+| 指定的 persona/透鏡不存在 | 列出可用選項供選擇 |
+| `--enhanced` 但宿主不支援子代理 | 靜默退回 baseline，照常進行 |
+| 前 3 名全來自同一 persona/透鏡 | 標示並在輸出前再跑一個透鏡 |
 
 ## References | 參考
 
 * [Brainstorm Assistant Skill](../brainstorm-assistant/SKILL.md)
+* [Brainstorm Assistant Guide](../brainstorm-assistant/guide.md)
 * [Requirement Assistant](../requirement-assistant/SKILL.md)
 * [Spec-Driven Development](../spec-driven-dev/SKILL.md)

package/bundled/skills/knowledge-graph/SKILL.md

@@ -14,7 +14,7 @@ Answer structural questions across specs, decisions, and code — *"what is the
 
 回答橫跨規格、決策與程式碼的結構性問題——*「XSPEC-205 的完整影響鏈是什麼？」*——依據[知識圖記憶標準](../../core/knowledge-graph-memory.md)的關係 schema。有無圖引擎皆可運作。
 
-> **Implements**: XSPEC-237 Phase 5 — knowledge-graph skill (CodeSage opt-in)
+> **Implements**: XSPEC-237 Phase 5 — knowledge-graph skill (EngramGraph opt-in)
 
 ## Mode Selection | 模式選擇
 
@@ -22,7 +22,7 @@ Detect which mode to use **before** answering:
 
 | Condition | Mode |
 |-----------|------|
-| `CODESAGE_URL` set, or a local graph engine responds on `/health` | Service mode (engine) |
+| `ENGRAM_URL` set, or a local graph engine responds on `/health` | Service mode (engine) |
 | Otherwise | Degraded mode (Markdown) |
 
 ## Workflow | 工作流程
@@ -31,7 +31,7 @@ Detect which mode to use **before** answering:
 2. **Choose mode** — probe for a graph engine (service) else fall back (degraded).
 3. **Service mode (AC-5b)** — issue a single multi-hop query and present the returned chain, including cross-domain links (code → spec → decision):
    ```bash
-   curl -s -X POST "$CODESAGE_URL/graph/impact-analysis" \
+   curl -s -X POST "$ENGRAM_URL/graph/impact-analysis" \
      -H 'content-type: application/json' \
      -d '{"nodeId":"XSPEC-205","maxHops":3}'
    ```
@@ -47,12 +47,12 @@ See [knowledge-graph-memory](../../core/knowledge-graph-memory.md). Front-matter
 
 ## Next Steps Guidance | 下一步引導
 
-- If degraded mode hit a reading-depth limit, tell the user a graph engine (e.g. CodeSage) would give a complete chain, and how to set `CODESAGE_URL`.
+- If degraded mode hit a reading-depth limit, tell the user a graph engine (e.g. EngramGraph) would give a complete chain, and how to set `ENGRAM_URL`.
 - If a referenced id was not found, surface it as a dangling reference to fix.
 - Offer to add missing `impacts`/`impacted_by` front-matter to the documents you traversed.
 
 ## Reference | 參考
 
 - Standard: [core/knowledge-graph-memory.md](../../core/knowledge-graph-memory.md)
-- Engine (opt-in): [CodeSage](https://github.com/AsiaOstrich/CodeSage) — `@asiaostrich/codesage`
+- Engine (opt-in): [EngramGraph](https://github.com/AsiaOstrich/EngramGraph) — `engramgraph`
 - Detailed guide: [guide.md](guide.md)

package/bundled/skills/knowledge-graph/guide.md

@@ -8,12 +8,12 @@ Companion to [SKILL.md](SKILL.md). Worked examples of both operating modes and t
 
 ## 1. Service Mode (graph engine) | 服務模式
 
-When a CodeSage-compatible engine is reachable, a single query returns the full chain.
+When an EngramGraph-compatible engine is reachable, a single query returns the full chain.
 
 ### Impact analysis | 影響分析
 
 ```bash
-curl -s -X POST "$CODESAGE_URL/graph/impact-analysis" \
+curl -s -X POST "$ENGRAM_URL/graph/impact-analysis" \
   -H 'content-type: application/json' \
   -d '{"nodeId":"XSPEC-205","maxHops":3}'
 # => { "nodeId": "XSPEC-205",
@@ -26,7 +26,7 @@ curl -s -X POST "$CODESAGE_URL/graph/impact-analysis" \
 ### Confidence feedback (SAGE) | 信心回饋
 
 ```bash
-curl -s -X POST "$CODESAGE_URL/graph/ingest" \
+curl -s -X POST "$ENGRAM_URL/graph/ingest" \
   -H 'content-type: application/json' \
   -d '{"type":"test_fail","functionId":"src/a.ts#execute"}'
 ```
@@ -66,4 +66,4 @@ Both modes produce the **same answer shape** (a list of connected Specs/Decision
 ## Reference | 參考
 
 - [core/knowledge-graph-memory.md](../../core/knowledge-graph-memory.md)
-- [CodeSage](https://github.com/AsiaOstrich/CodeSage)
+- [EngramGraph](https://github.com/AsiaOstrich/EngramGraph)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "universal-dev-standards",
-  "version": "5.15.1",
+  "version": "5.16.0",
   "description": "CLI tool for adopting Universal Development Standards",
   "keywords": [
     "documentation",
@@ -70,7 +70,7 @@
     "@vitest/coverage-v8": "^4.1.5",
     "ajv": "^8.20.0",
     "ajv-formats": "^3.0.1",
-    "eslint": "10.4.0",
+    "eslint": "10.4.1",
     "glob": "^13.0.1",
     "globals": "17.6.0",
     "husky": "^9.1.7",

package/src/commands/check.js

@@ -1861,13 +1861,14 @@ function displayWorkflowStatus(projectPath) {
  */
 async function runI18nLint(projectPath, options = {}) {
   const findings = lintI18nAll({ projectPath });
-  const { errors, warnings } = partitionI18nFindings(findings);
+  const { errors, warnings, infos } = partitionI18nFindings(findings);
 
   if (options.json) {
     console.log(JSON.stringify({
       summary: {
         errors: errors.length,
         warnings: warnings.length,
+        infos: infos.length,
       },
       findings,
     }, null, 2));
@@ -1900,8 +1901,16 @@ async function runI18nLint(projectPath, options = {}) {
     console.log();
   }
 
+  // Info findings are collapsed to a summary line to avoid flooding output
+  // (e.g. many locale files still lack source_hash). They never fail the gate.
+  if (infos.length > 0) {
+    console.log(chalk.cyan(`  ℹ ${infos.length} locale file(s) lack source_hash — silent content drift cannot be detected for them.`));
+    console.log(chalk.gray('    Stamp source_hash (sha256[:12] of canonical) when a locale is verified in sync to enable detection.'));
+    console.log();
+  }
+
   console.log(chalk.gray('─'.repeat(50)));
-  console.log(`  ${chalk.red('Errors:')} ${errors.length}    ${chalk.yellow('Warnings:')} ${warnings.length}`);
+  console.log(`  ${chalk.red('Errors:')} ${errors.length}    ${chalk.yellow('Warnings:')} ${warnings.length}    ${chalk.cyan('Info:')} ${infos.length}`);
   console.log();
 
   if (errors.length > 0) {

package/src/lint/i18n.js

@@ -8,7 +8,9 @@
  *   locale:description-must-match-language     (error)
  *   locale:must-have-source-frontmatter        (error)
  *   canonical:l3-language-consistency          (warn)
- *   translation-drift-warn                     (warn)
+ *   translation-drift-warn                     (warn)   version-gap (needs a version on the canonical)
+ *   translation-content-drift-warn             (warn)   source_hash mismatch — catches SILENT drift, version-independent (XSPEC-248)
+ *   translation-hash-missing                   (info)   blind-spot marker: no canonical version AND no source_hash → freshness unverifiable (XSPEC-248)
  *
  * `adopter-must-match-installed-locale` is deferred (needs project context,
  * not source-tree level).
@@ -18,6 +20,7 @@
 
 import { existsSync, readFileSync, readdirSync } from 'fs';
 import { join, dirname, basename, resolve } from 'path';
+import { createHash } from 'crypto';
 
 // CJK Unified Ideographs ranges (excluding fullwidth punctuation, but
 // including ranges Ext-A/B used by zh-TW/zh-CN). Hangul, Hiragana,
@@ -30,6 +33,11 @@ const CJK_REGEX = /[㐀-䶿一-鿿豈-﫿぀-ゟ゠-ヿ가-힯]/;
 // eslint-disable-next-line no-control-regex
 const NON_ASCII_REGEX = /[^\x09\x0A\x0D\x20-\x7E]/;
 
+// Global variant of CJK_REGEX (same character ranges, derived to avoid
+// re-typing the ranges) — used to *count* CJK characters for the
+// l3-language-consistency ratio check (bilingual templates vs. real drift).
+const CJK_GLOBAL_REGEX = new RegExp(CJK_REGEX.source, 'g');
+
 /**
  * Parse a minimal YAML front matter from markdown content.
  * Returns { fm: {key→value}, fmEndLine: number, body: string }
@@ -112,6 +120,26 @@ function readSourceVersion(filePath) {
   return null;
 }
 
+/**
+ * Compute a stable content hash of a canonical file: sha256 of its content
+ * with CRLF normalized and trailing whitespace stripped, truncated to 12 hex.
+ *
+ * Used for `source_hash` drift detection (XSPEC-248), which — unlike the
+ * version-gap check — does NOT depend on the canonical carrying a `version:`
+ * field. This is what catches "silent drift": canonical content changed but
+ * nobody bumped the locale's version metadata.
+ *
+ * @param {string} filePath - Absolute path to canonical file
+ * @returns {string|null} 12-hex digest, or null if file is missing
+ */
+export function computeSourceHash(filePath) {
+  if (!existsSync(filePath)) return null;
+  const normalized = readFileSync(filePath, 'utf-8')
+    .replace(/\r\n/g, '\n')
+    .replace(/\s+$/, '');
+  return createHash('sha256').update(normalized).digest('hex').slice(0, 12);
+}
+
 /**
  * Lint a canonical SKILL.md or core/*.md file for English-only frontmatter
  * and L3 language consistency.
@@ -119,14 +147,20 @@ function readSourceVersion(filePath) {
  * @param {string} skillMdPath - Absolute path to canonical file
  * @returns {Array<{rule, severity, line, message, file}>}
  */
-export function lintCanonical(skillMdPath) {
+export function lintCanonical(skillMdPath, opts = {}) {
+  // opts.isGuide — canonical guide.md (XSPEC-248): guide descriptions
+  // legitimately carry CJK discoverability keywords (e.g. "可觀測性"), so the
+  // ASCII-description rule does NOT apply. The l3 body check still does (and
+  // runs regardless of description language, since there is no ASCII rule to
+  // double-report against).
+  const { isGuide = false } = opts;
   const findings = [];
   if (!existsSync(skillMdPath)) return findings;
   const content = readFileSync(skillMdPath, 'utf-8');
   const { fm, body } = parseFrontmatter(content);
 
-  // Rule: canonical:description-must-be-ascii
-  if (fm && fm.description) {
+  // Rule: canonical:description-must-be-ascii (SKILL.md / core only; not guides)
+  if (!isGuide && fm && fm.description) {
     if (NON_ASCII_REGEX.test(fm.description.value)) {
       findings.push({
         rule: 'canonical:description-must-be-ascii',
@@ -140,9 +174,13 @@ export function lintCanonical(skillMdPath) {
 
   // Rule: canonical:l3-language-consistency
   // Heuristic: scan code-block-style output templates (```...```) for
-  // non-English example response text. Only fire when description is ASCII
-  // (otherwise the description-must-be-ascii error already covers it).
-  if (fm && fm.description && !NON_ASCII_REGEX.test(fm.description.value)) {
+  // non-English example response text. For SKILL.md/core only run when the
+  // description is ASCII (otherwise the description-must-be-ascii error already
+  // covers it); for guides (no ASCII rule) always scan the body.
+  const runL3 = isGuide
+    ? true
+    : Boolean(fm && fm.description && !NON_ASCII_REGEX.test(fm.description.value));
+  if (runL3) {
     const fenceRegex = /```([a-zA-Z0-9_-]*)\n([\s\S]*?)```/g;
     let match;
     while ((match = fenceRegex.exec(body)) !== null) {
@@ -158,19 +196,28 @@ export function lintCanonical(skillMdPath) {
       if (skipLangs.has(lang.toLowerCase())) continue;
 
       // Likely an output template / example response (e.g., ```markdown,
-      // ```text, ```output, ```response). Flag if it contains CJK.
+      // ```text, ```output, ```response). Only flag when the block is
+      // *predominantly* CJK — i.e. an untranslated example dumped into
+      // canonical. Deliberate bilingual `English | 中文` templates (house
+      // style) stay English-dominant, so they are allowed; a Chinese-only
+      // example response is CJK-dominant and still caught. (XSPEC-248 l3
+      // refinement — rule vs. bilingual-convention tension.)
       if (CJK_REGEX.test(block)) {
-        // Compute approx. line number of the fence start
-        const lineNum = content.substring(0, match.index).split('\n').length;
-        findings.push({
-          rule: 'canonical:l3-language-consistency',
-          severity: 'warn',
-          line: lineNum,
-          file: skillMdPath,
-          message: `Canonical body contains a non-English example response inside a non-code fenced block (lang=\`${lang || 'plain'}\`). Move translated examples to the locale variant.`
-        });
-        // Only report first occurrence per file to keep output readable
-        break;
+        const cjkCount = (block.match(CJK_GLOBAL_REGEX) || []).length;
+        const asciiLetterCount = (block.match(/[A-Za-z]/g) || []).length;
+        if (cjkCount > asciiLetterCount) {
+          // Compute approx. line number of the fence start
+          const lineNum = content.substring(0, match.index).split('\n').length;
+          findings.push({
+            rule: 'canonical:l3-language-consistency',
+            severity: 'warn',
+            line: lineNum,
+            file: skillMdPath,
+            message: `Canonical body contains a predominantly-CJK example response inside a non-code fenced block (lang=\`${lang || 'plain'}\`). Move the translated example to the locale variant (bilingual \`EN | 中文\` templates are allowed).`
+          });
+          // Only report first occurrence per file to keep output readable
+          break;
+        }
       }
     }
   }
@@ -233,12 +280,14 @@ export function lintLocale(skillMdPath, locale) {
     }
   }
 
-  // Rule: translation-drift-warn
+  // Rules: translation-drift-warn (version) + translation-content-drift-warn (hash) + translation-hash-missing
   if (fm.source && fm.translation_version) {
     // Resolve source path relative to this file's directory
     const sourceRel = fm.source.value;
     const sourcePath = resolve(dirname(skillMdPath), sourceRel);
     const actualSourceVersion = readSourceVersion(sourcePath);
+
+    // (a) version-gap drift — only works when the canonical carries a version
     if (actualSourceVersion) {
       const gap = isMajorOrMinorGap(fm.translation_version.value, actualSourceVersion);
       if (gap) {
@@ -254,14 +303,40 @@ export function lintLocale(skillMdPath, locale) {
         });
       }
     }
+
+    // (b) content-hash drift — version-independent; catches SILENT drift (XSPEC-248)
+    if (fm.source_hash) {
+      const currentHash = computeSourceHash(sourcePath);
+      if (currentHash && currentHash !== fm.source_hash.value) {
+        findings.push({
+          rule: 'translation-content-drift-warn',
+          severity: 'warn',
+          line: fm.source_hash.line,
+          file: skillMdPath,
+          message: `Canonical content changed since last sync (source_hash \`${fm.source_hash.value}\` != current \`${currentHash}\`). Re-translate and update source_hash.`
+        });
+      }
+    } else if (!actualSourceVersion) {
+      // (c) blind-spot marker: neither a canonical version NOR a source_hash exists,
+      // so content freshness cannot be verified at all (how brainstorm rotted to v1.0).
+      findings.push({
+        rule: 'translation-hash-missing',
+        severity: 'info',
+        line: fm.source.line,
+        file: skillMdPath,
+        message: 'Cannot verify content freshness: canonical has no version field and locale has no source_hash. Stamp source_hash (sha256[:12] of canonical) to enable silent-drift detection.'
+      });
+    }
   }
 
   return findings;
 }
 
 /**
- * Batch-lint a UDS project tree. Scans canonical `skills/` and `core/`,
- * plus `locales/{locale}/skills/` and `locales/{locale}/core/`.
+ * Batch-lint a UDS project tree. Scans canonical `skills/` (SKILL.md + guide.md)
+ * and `core/`, plus `locales/{locale}/skills/` (SKILL.md + guide.md) and
+ * `locales/{locale}/core/`. Canonical guide.md uses the guide ruleset
+ * (CJK discoverability keywords allowed in description; XSPEC-248).
  *
  * @param {object} opts
  * @param {string} opts.projectPath - Project (or UDS) root path
@@ -276,7 +351,7 @@ export function lintAll({ projectPath, locales }) {
   const coreDir = join(projectPath, 'core');
   const localesDir = join(projectPath, 'locales');
 
-  // --- Canonical skills ---
+  // --- Canonical skills (SKILL.md + guide.md) ---
   if (existsSync(skillsDir)) {
     for (const entry of readdirSync(skillsDir, { withFileTypes: true })) {
       if (!entry.isDirectory()) continue;
@@ -284,6 +359,11 @@ export function lintAll({ projectPath, locales }) {
       if (existsSync(skillFile)) {
         findings.push(...lintCanonical(skillFile));
       }
+      // XSPEC-248: guide.md uses the guide ruleset (CJK description allowed).
+      const guideFile = join(skillsDir, entry.name, 'guide.md');
+      if (existsSync(guideFile)) {
+        findings.push(...lintCanonical(guideFile, { isGuide: true }));
+      }
     }
   }
 
@@ -313,6 +393,11 @@ export function lintAll({ projectPath, locales }) {
           if (existsSync(skillFile)) {
             findings.push(...lintLocale(skillFile, locale));
           }
+          // XSPEC-248: locale guide.md reuses the locale ruleset (freshness checks).
+          const guideFile = join(localeSkillsDir, entry.name, 'guide.md');
+          if (existsSync(guideFile)) {
+            findings.push(...lintLocale(guideFile, locale));
+          }
         }
       }
       if (existsSync(localeCoreDir)) {
@@ -334,5 +419,6 @@ export function partitionFindings(findings) {
   return {
     errors: findings.filter(f => f.severity === 'error'),
     warnings: findings.filter(f => f.severity === 'warn'),
+    infos: findings.filter(f => f.severity === 'info'),
   };
 }

package/standards-registry.json

@@ -1,6 +1,6 @@
 {
   "$schema": "https://json-schema.org/draft/2020-12/schema",
-  "version": "5.15.1",
+  "version": "5.16.0",
   "lastUpdated": "2026-05-13",
   "description": "Standards registry for universal-dev-standards with integrated skills and AI-optimized formats",
   "formats": {
@@ -58,14 +58,14 @@
     "standards": {
       "name": "universal-dev-standards",
       "url": "https://github.com/AsiaOstrich/universal-dev-standards",
-      "version": "5.15.1"
+      "version": "5.16.0"
     },
     "skills": {
       "name": "universal-dev-standards",
       "url": "https://github.com/AsiaOstrich/universal-dev-standards",
       "localPath": "skills",
       "rawUrl": "https://raw.githubusercontent.com/AsiaOstrich/universal-dev-standards/main/skills",
-      "version": "5.15.1",
+      "version": "5.16.0",
       "note": "Skills are now included in the main repository under skills/"
     }
   },
@@ -2350,7 +2350,7 @@
       "id": "license-compliance",
       "name": "License Compliance Standards",
       "nameZh": "授權合規標準",
-      "version": "5.15.1",
+      "version": "5.16.0",
       "source": {
         "human": "core/license-compliance.md",
         "ai": "ai/standards/license-compliance.ai.yaml"