@dommaker/harness 0.16.6 → 0.16.8

package/src/__tests__/doctor-evolver.test.ts

@@ -9,9 +9,9 @@ import { TraceCollector } from '../monitoring/traces';
 import { TraceAnalyzer } from '../monitoring/trace-analyzer';
 import { ConstraintDoctor, createDoctor } from '../monitoring/constraint-doctor';
 import { ConstraintEvolver, createEvolver } from '../monitoring/constraint-evolver';
+import { getConstraint } from '../core/constraints/definitions';
 import type { ExecutionTrace, TraceAnomaly } from '../types/trace';
-import type { Diagnosis } from '../monitoring/constraint-doctor';
-import type { ConstraintProposal } from '../monitoring/constraint-evolver';
+import type { Diagnosis, ConstraintProposal } from '../types/monitoring-types';
 
 describe('ConstraintDoctor', () => {
   let tempDir: string;
@@ -174,7 +174,7 @@ describe('ConstraintEvolver', () => {
   beforeEach(() => {
     tempDir = path.join(os.tmpdir(), `harness-proposals-${Date.now()}`);
     fs.mkdirSync(tempDir, { recursive: true });
-    evolver = createEvolver(tempDir);
+    evolver = createEvolver(tempDir, getConstraint);
     doctor = createDoctor();
   });
 

package/src/__tests__/iron-laws.test.ts

@@ -217,30 +217,3 @@ describe('Constraint Levels', () => {
     });
   });
 });
-
-describe('IronLawViolationError (deprecated)', () => {
-  it('should be instanceof ConstraintViolationError', () => {
-    const { IronLawViolationError, ConstraintViolationError } = require('../types/constraint');
-    
-    const result = {
-      id: 'test',
-      level: 'iron_law',
-      satisfied: false,
-      constraint: {
-        id: 'test',
-        level: 'iron_law',
-        rule: 'TEST',
-        message: 'test',
-        trigger: 'code_implementation',
-        enforcement: 'test',
-      },
-      message: 'Test error',
-      checkedAt: new Date(),
-    };
-
-    const error = new IronLawViolationError(result);
-    expect(error).toBeInstanceOf(ConstraintViolationError);
-    expect(error.name).toBe('IronLawViolationError');
-    expect(error.result).toBe(result);
-  });
-});

package/src/__tests__/presets.test.ts

@@ -3,7 +3,7 @@
  */
 
 import { describe, it, expect } from '@jest/globals';
-import { STRICT_PRESET, STANDARD_PRESET, RELAXED_PRESET, getPreset, getIronLawPreset, STANDARD_IRON_LAWS_CONFIG } from '../presets/standard';
+import { STRICT_PRESET, STANDARD_PRESET, RELAXED_PRESET, getPreset } from '../presets/standard';
 
 describe('Presets', () => {
   describe('STRICT_PRESET', () => {
@@ -62,43 +62,4 @@ describe('Presets', () => {
       expect(preset.name).toBe('standard');
     });
   });
-
-  describe('getIronLawPreset (deprecated)', () => {
-    it('strict 应该返回所有 Iron Laws', () => {
-      const config = getIronLawPreset('strict');
-      expect(config.preset).toBe('strict');
-      expect(config.enabled).toBe(true);
-      expect(Array.isArray(config.ironLaws)).toBe(true);
-      expect(config.ironLaws.length).toBeGreaterThan(0);
-    });
-
-    it('standard 应该返回所有 Iron Laws', () => {
-      const config = getIronLawPreset('standard');
-      expect(config.preset).toBe('standard');
-      expect(config.enabled).toBe(true);
-      expect(Array.isArray(config.ironLaws)).toBe(true);
-      expect(config.ironLaws.length).toBeGreaterThan(0);
-    });
-
-    it('relaxed 应该返回过滤后的 Iron Laws', () => {
-      const config = getIronLawPreset('relaxed');
-      expect(config.preset).toBe('relaxed');
-      expect(config.enabled).toBe(true);
-      expect(Array.isArray(config.ironLaws)).toBe(true);
-      expect(config.ironLaws.length).toBe(3);
-    });
-
-    it('未知预设应该返回所有 Iron Laws', () => {
-      const config = getIronLawPreset('unknown');
-      expect(config.preset).toBe('unknown');
-      expect(config.enabled).toBe(true);
-      expect(config.ironLaws.length).toBeGreaterThan(0);
-    });
-  });
-
-  describe('STANDARD_IRON_LAWS_CONFIG (deprecated re-export)', () => {
-    it('应该等同于 STANDARD_PRESET', () => {
-      expect(STANDARD_IRON_LAWS_CONFIG).toBe(STANDARD_PRESET);
-    });
-  });
 });

package/src/architecture/constraint-engine.ts

@@ -289,7 +289,6 @@ function parseSimpleYaml(content: string): { rules?: ArchitectureRule[] } {
   const result: { rules?: ArchitectureRule[] } = { rules: [] };
   let currentRule: Partial<ArchitectureRule> | null = null;
   let currentArray: string[] | null = null;
-  let indentLevel = 0;
   
   for (const line of lines) {
     const trimmed = line.trim();
@@ -297,9 +296,6 @@ function parseSimpleYaml(content: string): { rules?: ArchitectureRule[] } {
     // 跳过注释和空行
     if (!trimmed || trimmed.startsWith('#')) continue;
     
-    // 检测缩进级别
-    const leadingSpace = line.length - line.trimStart().length;
-    
     // 新规则开始
     if (trimmed.startsWith('- id:')) {
       if (currentRule && result.rules) {
@@ -307,7 +303,6 @@ function parseSimpleYaml(content: string): { rules?: ArchitectureRule[] } {
       }
       currentRule = { id: trimmed.replace('- id:', '').trim() };
       currentArray = null;
-      indentLevel = leadingSpace;
     }
     // 规则属性
     else if (currentRule && trimmed.includes(':')) {

package/src/architecture/cross-project-checker.ts

@@ -8,7 +8,8 @@
  */
 
 import { runCommand } from '../utils/exec';
-import { existsSync, readFileSync, readdirSync } from 'fs';
+import { findTsSourceFiles } from '../utils/file-walk';
+import { existsSync, readFileSync } from 'fs';
 import * as path from 'path';
 
 export interface CrossProjectViolation {
@@ -166,7 +167,7 @@ async function checkTypeConsistency(
 ): Promise<CrossProjectViolation[]> {
   const violations: CrossProjectViolation[] = [];
 
-  for (const [pair, locations] of Object.entries(config.contractLocations)) {
+  for (const [pair] of Object.entries(config.contractLocations)) {
     const [from, to] = pair.split('-');
     if (!context.changedProjects.includes(from) && !context.changedProjects.includes(to)) {
       continue;
@@ -334,25 +335,6 @@ interface ApiChange {
 
 // ============ 辅助函数实现 ============
 
-/**
- * 递归查找目录中的 .ts 文件（排除 node_modules/__tests__/dist/.d.ts）
- */
-function findTsFiles(dir: string): string[] {
-  if (!existsSync(dir)) return [];
-  const results: string[] = [];
-  const entries = readdirSync(dir, { withFileTypes: true });
-  for (const entry of entries) {
-    const fullPath = path.join(dir, entry.name);
-    if (entry.isDirectory()) {
-      if (entry.name === 'node_modules' || entry.name === '__tests__' || entry.name === 'dist') continue;
-      results.push(...findTsFiles(fullPath));
-    } else if (entry.name.endsWith('.ts') && !entry.name.endsWith('.d.ts')) {
-      results.push(fullPath);
-    }
-  }
-  return results;
-}
-
 function extractApiName(file: string): string {
   return path.basename(file, '.ts');
 }
@@ -388,7 +370,7 @@ async function detectChangeType(file: string, baseBranch: string): Promise<ApiCh
 function extractExportedTypes(project: string): string[] {
   const projectPath = resolveProjectPath(project);
   const srcDir = path.join(projectPath, 'src');
-  const files = findTsFiles(srcDir);
+  const files = findTsSourceFiles(srcDir);
   const types = new Set<string>();
 
   for (const file of files) {
@@ -410,7 +392,7 @@ function extractExportedTypes(project: string): string[] {
 function extractTypeUsage(project: string, types: string[]): Record<string, { hasMismatch: boolean; expected?: string; actual?: string }> {
   const projectPath = resolveProjectPath(project);
   const srcDir = path.join(projectPath, 'src');
-  const files = findTsFiles(srcDir);
+  const files = findTsSourceFiles(srcDir);
   const result: Record<string, { hasMismatch: boolean; expected?: string; actual?: string }> = {};
 
   for (const typeName of types) {
@@ -487,7 +469,7 @@ function parseDocInterfaces(docPath: string): Record<string, Array<{name: string
 function extractExportedApis(project: string): Array<{name: string, signature: string}> {
   const projectPath = resolveProjectPath(project);
   const srcDir = path.join(projectPath, 'src');
-  const files = findTsFiles(srcDir);
+  const files = findTsSourceFiles(srcDir);
   const apis: Array<{name: string, signature: string}> = [];
   const seen = new Set<string>();
 

package/src/cli/commands/__tests__/check.test.ts

@@ -28,7 +28,8 @@ jest.mock('child_process', () => ({
   exec: jest.fn(),
   execSync: jest.fn((cmd: string) => {
     if (cmd.includes('git ls-tree')) {
-      return Buffer.from('040000 tree abc123\tsrc\n'); // directory exists in HEAD
+      // 工单 18 后为批量命令 `git ls-tree -r --name-only HEAD`，返回文件列表
+      return Buffer.from('src/foo.ts\nsrc/module.ts\nsrc/__tests__/module.test.ts\n');
     }
     if (cmd.includes('git diff')) {
       return Buffer.from('src/test.ts\nsrc/legacy.js');
@@ -41,6 +42,7 @@ jest.mock('child_process', () => ({
 jest.mock('../../../core/constraints/checker', () => ({
   constraintChecker: {
     setCustomConfig: jest.fn(),
+    setTraceRecorder: jest.fn(),
     checkConstraints: jest.fn(),
   },
 }));

package/src/cli/commands/__tests__/sync-docs.test.ts

@@ -169,6 +169,53 @@ describe('sync-docs command', () => {
       fs.rmSync(testDir, { recursive: true, force: true });
     });
 
+    it('描述以 / 结尾的表格行不应被误判为目录条目（check 必须收敛）', async () => {
+      // 回归：JSDoc 首行在 "gitRepo /" 处换行时，生成的描述单元格以 / 结尾，
+      // 旧目录条目正则将其误判为目录 → --check 永远报「包含已删除的模块」
+      // （2026-08-04 studio PR #44 CI Governance 不收敛事故）
+      const testDir = path.join(tempDir, 'trailing-slash-desc');
+      const srcDir = path.join(testDir, 'src');
+      fs.mkdirSync(srcDir, { recursive: true });
+
+      fs.writeFileSync(
+        path.join(srcDir, 'workspace.ts'),
+        '/**\n * 归属链解析 — metadata.workspaceRoot（Requirement→gitRepo /\n * 人工绑定）优先。\n */\nexport const x = 1;\n'
+      );
+
+      // 第一轮：生成 CAPABILITIES.md（描述行以 / 结尾）
+      await syncDocs({ projectPath: testDir });
+      const cap = fs.readFileSync(path.join(testDir, 'CAPABILITIES.md'), 'utf-8');
+      expect(cap).toContain('workspace.ts');
+      expect(cap).toMatch(/gitRepo\s*\/\s*\|/);
+
+      // 第二轮：生成后立即 --check 必须收敛（修复前此处必 false）
+      consoleSpy.mockClear();
+      const check = await syncDocs({ projectPath: testDir, check: true });
+      expect(check).toBe(true);
+
+      fs.rmSync(testDir, { recursive: true, force: true });
+    });
+
+    it('真正的目录条目（如 agents/）仍应被识别', async () => {
+      const testDir = path.join(tempDir, 'dir-entry');
+      const srcDir = path.join(testDir, 'src');
+      fs.mkdirSync(srcDir, { recursive: true });
+
+      fs.writeFileSync(path.join(srcDir, 'mymodule.ts'), 'export const x = 1;');
+
+      // CAPABILITIES.md 用目录条目记录 src/ 下的模块
+      fs.writeFileSync(
+        path.join(testDir, 'CAPABILITIES.md'),
+        '# Capabilities\n\n| 模块 | 文件 | 说明 |\n|------|------|------|\n| all | src/ | 全量记录 |'
+      );
+
+      // src/ 目录条目覆盖 mymodule.ts → 无 added，check 通过
+      const result = await syncDocs({ projectPath: testDir, check: true });
+      expect(result).toBe(true);
+
+      fs.rmSync(testDir, { recursive: true, force: true });
+    });
+
     it('应该检查 CONTEXT.md 是否存在', async () => {
       const testDir = path.join(tempDir, 'context-check');
       const harnessDir = path.join(testDir, '.harness');
@@ -358,5 +405,35 @@ describe('sync-docs command', () => {
 
       fs.rmSync(testDir, { recursive: true, force: true });
     });
+
+    it('basename 碰撞的幽灵条目（文件已删但同名文件存在）也应被移除', async () => {
+      const testDir = path.join(tempDir, 'ghost-basename-collision');
+      const srcA = path.join(testDir, 'src', 'a');
+      fs.mkdirSync(srcA, { recursive: true });
+
+      // 真实存在的同名文件
+      fs.writeFileSync(path.join(srcA, 'routes.ts'), 'export const routes = 1;');
+
+      // CAPABILITIES.md 另有一条指向不存在路径的 routes.ts（basename 碰撞，
+      // 按 basename 对比时永远不可见 —— 2026-08-08 studio CI 4 连红事故）
+      fs.writeFileSync(
+        path.join(testDir, 'CAPABILITIES.md'),
+        [
+          '# CAPABILITIES.md', '',
+          '> 最后更新: 2026-01-01', '', '---', '',
+          '| 模块 | 文件 | 说明 |', '|------|------|------|',
+          '| routes | src/a/routes.ts | 真实文件 |',
+          '| routes | src/ghost/routes.ts | 幽灵条目 |',
+          '',
+        ].join('\n'),
+      );
+
+      await syncDocs({ projectPath: testDir });
+      const content = fs.readFileSync(path.join(testDir, 'CAPABILITIES.md'), 'utf-8');
+      expect(content).toContain('src/a/routes.ts');
+      expect(content).not.toContain('src/ghost/routes.ts');
+
+      fs.rmSync(testDir, { recursive: true, force: true });
+    });
   });
 });

package/src/cli/commands/__tests__/validate.test.ts

@@ -96,8 +96,13 @@ describe('validate command', () => {
       };
       (MockCheckpointValidator.getInstance as jest.Mock).mockReturnValue(mockValidator);
 
+      // 工单 23：检查点失败一律 exit 1（门控语义）
+      const mockExit = jest.spyOn(process, 'exit').mockImplementation(() => undefined as never);
+
       await validate({});
       expect(consoleSpy).toHaveBeenCalledWith(expect.stringContaining('失败'));
+      expect(mockExit).toHaveBeenCalledWith(1);
+      mockExit.mockRestore();
     });
 
     it('应该在严格模式下退出', async () => {

package/src/cli/commands/analyze-sessions.ts

@@ -7,12 +7,27 @@
  *   3. 规则缺口（高频模式未被现有 memory 覆盖）
  *
  * 输出候选规则建议，供用户审核后写入 memory。
+ *
+ * 工单 19-C：transcript 解析/纠正模式/分词相似度收敛至 cli/session-mining/。
  */
 
 import chalk from 'chalk';
 import * as fs from 'fs';
 import * as path from 'path';
 import * as os from 'os';
+import {
+  readTranscriptSessions,
+  type MinedSession,
+  extractCorrectionMatches,
+  cleanCorrectionConcept,
+  tokenize,
+  stripCodeBlocks,
+  jaccardSimilarity,
+  isPunctuation,
+  isCodeNoise,
+  hasSemanticContent,
+  STOP_WORDS,
+} from '../session-mining';
 
 export interface AnalyzeSessionsOptions {
   days?: number;
@@ -35,38 +50,6 @@ interface PatternCandidate {
   suggestedRule: string;      // 建议的规则名
 }
 
-// ── Correction pattern templates ──
-// Match user sentences that indicate repeating a request
-const CORRECTION_PATTERNS = [
-  /你(?:又|还是)(?:在)?(?:犯|忘|没|不).{0,20}?[了]?/g,
-  /我不是(?:说|让|让做|讲)[了过]?.{0,30}?[了吗?]?/g,
-  /怎么(?:又|还|老)(?:是|在)?.{0,20}?[了]?/g,
-  /我(?:一直|反复|总是)(?:在|说|强调)?.{0,20}?[了]?/g,
-  /老(?:是|在|犯|忘)(?:了)?.{0,20}?[了]?/g,
-  /这(?:个|种)(?:问题|模式|错误).{0,20}?[.。！]?/g,
-  /(?:第[一二三四五六七八九十\d]+次|反复|重复)(?:说|提醒|强调).{0,20}?/g,
-  /不(?:要|能|想|愿意).{0,20}?老.{0,10}?[了]?/g,
-  /(?:补上|加上|记上|修复).{0,10}?[了吗?？]?/g,
-  /(?:沉淀|监控|日志|记录).{0,5}?[了吗?？]?/g,
-];
-
-// ── N-gram stop words ──
-const STOP_WORDS = new Set([
-  '的', '了', '在', '是', '我', '有', '和', '就', '不', '人', '都', '一',
-  '一个', '上', '也', '很', '到', '说', '要', '去', '你', '会', '着',
-  '没有', '看', '好', '自己', '这', '他', '她', '它', '们', '那',
-  'the', 'a', 'an', 'is', 'are', 'was', 'were', 'be', 'been', 'being',
-  'have', 'has', 'had', 'do', 'does', 'did', 'will', 'would', 'could',
-  'should', 'may', 'might', 'can', 'shall', 'to', 'of', 'in', 'for',
-  'on', 'with', 'at', 'by', 'from', 'as', 'into', 'through', 'during',
-  'before', 'after', 'above', 'below', 'between', 'under', 'again',
-  'further', 'then', 'once', 'here', 'there', 'when', 'where', 'why',
-  'how', 'all', 'both', 'each', 'few', 'more', 'most', 'other', 'some',
-  'such', 'only', 'own', 'same', 'so', 'than', 'too', 'very', 'just',
-  '因为', '所以', '但是', '而且', '如果', '虽然', '可以', '这个', '那个',
-  '什么', '怎么', '哪', '吗', '吧', '呢', '啊', '嗯', '哦',
-]);
-
 // ── Main ──
 
 export async function analyzeSessions(options: AnalyzeSessionsOptions): Promise<void> {
@@ -80,9 +63,11 @@ export async function analyzeSessions(options: AnalyzeSessionsOptions): Promise<
     return;
   }
 
-  // 1. Scan transcripts
+  // 1. Scan transcripts（按修改时间过滤 + 倒序）
   const cutoff = Date.now() - days * 24 * 60 * 60 * 1000;
-  const sessions = findSessions(transcriptDir, cutoff);
+  const sessions = readTranscriptSessions(transcriptDir)
+    .filter(s => s.mtimeMs >= cutoff)
+    .sort((a, b) => b.date.localeCompare(a.date));
 
   if (sessions.length === 0) {
     console.log(chalk.yellow(`No sessions found in the last ${days} days`));
@@ -108,62 +93,12 @@ export async function analyzeSessions(options: AnalyzeSessionsOptions): Promise<
   console.log(chalk.blue(`🔍 Analyzing ${sessions.length} sessions (last ${days} days)...\n`));
 
   printCorrectionSummary(corrections);
-  printCandidates(candidates, existingRules);
-}
-
-// ── Scan ──
-
-interface Session {
-  id: string;
-  date: string;
-  turns: Array<{ role: string; content: string }>;
-}
-
-function findSessions(dir: string, cutoff: number): Session[] {
-  const sessions: Session[] = [];
-  try {
-    for (const file of fs.readdirSync(dir)) {
-      if (!file.endsWith('.jsonl')) continue;
-      const filePath = path.join(dir, file);
-      const stat = fs.statSync(filePath);
-      if (stat.mtimeMs < cutoff) continue;
-
-      const turns: Array<{ role: string; content: string }> = [];
-      const content = fs.readFileSync(filePath, 'utf-8');
-      for (const line of content.split('\n')) {
-        if (!line.trim()) continue;
-        try {
-          const entry = JSON.parse(line);
-          // Claude Code transcript format: { type: "user"|"assistant", message: { role, content } }
-          const msg = entry.message;
-          if (msg?.role && (msg.content || entry.type === 'user')) {
-            const text = typeof msg.content === 'string'
-              ? msg.content
-              : Array.isArray(msg.content)
-                ? msg.content.filter((b: any) => b.type === 'text').map((b: any) => b.text).join(' ')
-                : '';
-            turns.push({ role: msg.role, content: text });
-          }
-        } catch {}
-      }
-
-      if (turns.length > 0) {
-        sessions.push({
-          id: file.replace('.jsonl', '').slice(0, 40),
-          date: stat.mtime.toISOString().slice(0, 10),
-          turns,
-        });
-      }
-    }
-  } catch (e) {
-    // Directory might not exist
-  }
-  return sessions.sort((a, b) => b.date.localeCompare(a.date));
+  printCandidates(candidates);
 }
 
 // ── Correction Extraction ──
 
-function extractCorrections(sessions: Session[]): Correction[] {
+function extractCorrections(sessions: MinedSession[]): Correction[] {
   const corrections: Correction[] = [];
 
   for (const session of sessions) {
@@ -171,18 +106,13 @@ function extractCorrections(sessions: Session[]): Correction[] {
       if (turn.role !== 'user') continue;
       const text = stripCodeBlocks(turn.content);
 
-      for (const pattern of CORRECTION_PATTERNS) {
-        pattern.lastIndex = 0;
-        let match;
-        while ((match = pattern.exec(text)) !== null) {
-          const sentence = match[0].trim();
-          if (sentence.length > 3 && sentence.length < 200) {
-            corrections.push({
-              sentence,
-              sessionId: session.id,
-              timestamp: session.date,
-            });
-          }
+      for (const sentence of extractCorrectionMatches(text)) {
+        if (sentence.length > 3 && sentence.length < 200) {
+          corrections.push({
+            sentence,
+            sessionId: session.id,
+            timestamp: session.date,
+          });
         }
       }
     }
@@ -205,7 +135,7 @@ interface NGramResult {
   sessions: string[];
 }
 
-function extractNGrams(sessions: Session[], opts: NGramOptions): NGramResult[] {
+function extractNGrams(sessions: MinedSession[], opts: NGramOptions): NGramResult[] {
   const phraseMap = new Map<string, { count: number; sessions: Set<string> }>();
 
   for (const session of sessions) {
@@ -244,42 +174,6 @@ function extractNGrams(sessions: Session[], opts: NGramOptions): NGramResult[] {
     .sort((a, b) => b.frequency - a.frequency || b.sessions.length - a.sessions.length);
 }
 
-function tokenize(text: string): string[] {
-  const tokens: string[] = [];
-  let current = '';
-  const CN = /[\u4e00-\u9fff\u3400-\u4dbf]/;
-
-  for (const char of text) {
-    if (CN.test(char)) {
-      current += char;
-    } else if (/[a-zA-Z0-9_\-/.]/.test(char)) {
-      current += char;
-    } else {
-      pushToken(current, tokens, CN);
-      current = '';
-    }
-  }
-  pushToken(current, tokens, CN);
-
-  return tokens;
-}
-
-function pushToken(current: string, tokens: string[], CN: RegExp): void {
-  if (current.length < 2) return;
-  if (CN.test(current)) {
-    // Chinese: character bigrams + trigrams + the full sequence
-    for (let i = 0; i <= current.length - 2; i++) {
-      tokens.push(current.slice(i, i + 2));
-    }
-    for (let i = 0; i <= current.length - 3; i++) {
-      tokens.push(current.slice(i, i + 3));
-    }
-    if (current.length <= 6) tokens.push(current); // keep short full sequences too
-  } else {
-    tokens.push(current);
-  }
-}
-
 // ── Existing Rules ──
 
 function isInExistingRules(concept: string, existingRules: Set<string>): boolean {
@@ -401,16 +295,7 @@ function clusterCorrections(corrections: Correction[]): CorrectionCluster[] {
 function extractConcept(sentences: string[]): string {
   // Take the longest sentence and extract the noun phrase after the correction marker
   const longest = sentences.reduce((a, b) => a.length > b.length ? a : b, '');
-  // Remove correction patterns, keep the concept
-  const cleaned = longest
-    .replace(/你(?:又|还是)(?:在)?(?:犯|忘|没|不)/g, '')
-    .replace(/我不是(?:说|让|让做|讲)[了过]?/g, '')
-    .replace(/怎么(?:又|还|老)(?:是|在)?/g, '')
-    .replace(/我(?:一直|反复|总是)(?:在|说|强调)?/g, '')
-    .replace(/老(?:是|在|犯|忘)(?:了)?/g, '')
-    .replace(/[，。！？、；：""''（）\s]+/g, '')
-    .trim();
-
+  const cleaned = cleanCorrectionConcept(longest);
   return cleaned.slice(0, 60) || longest.slice(0, 60);
 }
 
@@ -425,7 +310,7 @@ function printCorrectionSummary(corrections: Correction[]): void {
   console.log(chalk.yellow(`📢 ${corrections.length} correction signals detected\n`));
 }
 
-function printCandidates(candidates: PatternCandidate[], existingRules: Set<string>): void {
+function printCandidates(candidates: PatternCandidate[]): void {
   if (candidates.length === 0) {
     console.log(chalk.green('✅ No new pattern candidates — all recurring concepts already have rules\n'));
     return;
@@ -447,42 +332,3 @@ function printCandidates(candidates: PatternCandidate[], existingRules: Set<stri
     console.log(chalk.gray(`... and ${candidates.length - 10} more candidates. Run with --json for full list.`));
   }
 }
-
-// ── Utilities ──
-
-function stripCodeBlocks(text: string): string {
-  return text
-    .replace(/```[\s\S]*?```/g, '')
-    .replace(/`[^`]+`/g, '')
-    .replace(/\{[^{}]*"[^"]+"\s*:\s*"[^"]*"[^}]*\}/g, '')  // strip JSON objects
-    .replace(/\[[\s\S]*?\]/g, '')  // strip JSON arrays
-    .replace(/\\n/g, ' ')  // strip escaped newlines
-    .replace(/[{}[\]"':,\\]+/g, ' '); // strip JSON punctuation
-}
-
-function hasSemanticContent(phrase: string): boolean {
-  // Must contain Chinese characters or be a meaningful English word (≥7 chars)
-  if (/[\u4e00-\u9fff]/.test(phrase)) return true;
-  if (phrase.length >= 7 && /^[a-zA-Z][a-zA-Z0-9_]*[a-zA-Z]$/.test(phrase)) return true;
-  return false;
-}
-
-function isCodeNoise(phrase: string): boolean {
-  // Pure Latin chars without Chinese → likely JSON/code artifact
-  if (!/[\u4e00-\u9fff]/.test(phrase) && phrase.length <= 6) return true;
-  // Common JSON field name fragments
-  if (/^(?:clla|cood|odde|oppe|laaw|peen|enne|ennc|ddec|deco|ecod|code|type|role|name|text|file|path|tool|user|session)$/i.test(phrase)) return true;
-  return false;
-}
-
-function isPunctuation(str: string): boolean {
-  return /^[，。！？、；：""''（）\[\]【】「」『』《》〈〉\s]+$/.test(str);
-}
-
-function jaccardSimilarity(a: string, b: string): number {
-  const setA = new Set(tokenize(a));
-  const setB = new Set(tokenize(b));
-  const intersection = new Set([...setA].filter(x => setB.has(x)));
-  const union = new Set([...setA, ...setB]);
-  return union.size === 0 ? 0 : intersection.size / union.size;
-}