@su-record/vibe 2.9.18 → 2.9.19

package/hooks/scripts/prompt-dispatcher.js

@@ -1,190 +1,190 @@
-#!/usr/bin/env node
-/**
- * UserPromptSubmit 디스패처
- *
- * UserPromptSubmit은 matcher를 지원하지 않아 모든 hook이 매번 실행됨.
- * 이 디스패처가 stdin에서 prompt를 읽고, 패턴 매칭 후 해당 스크립트만 실행.
- *
- * 이점:
- * - 외부 LLM 호출(GPT/Gemini)이 패턴 매칭 없이 발동하지 않음
- * - context window에 불필요한 응답이 주입되지 않음
- * - 단일 프로세스에서 매칭 후 필요한 스크립트만 fork
- */
-import { execFile } from 'child_process';
-import { fileURLToPath } from 'url';
-import path from 'path';
-
-const __dirname = path.dirname(fileURLToPath(import.meta.url));
-
-// 재귀 가드 — 자식 Claude 세션에서 이 hook이 다시 실행되는 것 차단.
-// llm-orchestrate.js의 callClaudeCli가 VIBE_HOOK_DEPTH=1을 주입하므로,
-// 값이 있으면 즉시 종료해 프로세스 폭탄을 막는다.
-if (process.env.VIBE_HOOK_DEPTH) process.exit(0);
-
-// stdin에서 prompt 읽기
-let inputData = '';
-for await (const chunk of process.stdin) {
-  inputData += chunk;
-}
-
-let prompt = '';
-try {
-  const parsed = JSON.parse(inputData);
-  prompt = parsed.prompt || '';
-} catch {
-  process.exit(0);
-}
-
-if (!prompt) process.exit(0);
-
-// 패턴 → 실행할 스크립트 매핑
-// 각 항목: { pattern, script, args, label }
-const DISPATCH_RULES = [
-  // 항상 실행 (경량 스크립트)
-  {
-    pattern: null, // always
-    script: 'keyword-detector.js',
-    args: [prompt],
-    label: 'keyword',
-  },
-
-  // 패턴 매칭이 필요한 스크립트
-  {
-    pattern: /ultrawork|ulw|울트라워크|ralph|ralplan/i,
-    script: null, // keyword-detector가 이미 처리
-    label: 'skip',
-  },
-  // echo 전용 (stdout으로 직접 출력)
-  {
-    pattern: /e2e.*테스트|e2e.*test|playwright|브라우저.*테스트|browser.*test/i,
-    script: null,
-    echo: '[E2E MODE] Use /vibe.utils --e2e for Playwright-based browser testing. Supports visual regression and video recording.',
-    label: 'e2e-echo',
-  },
-
-  // 외부 LLM 호출 (GPT/Gemini) - 패턴 매칭 필수
-  {
-    pattern: /아키텍처.*(검토|리뷰|분석)|architecture.*(review|analyz)|설계.*검토|구조.*분석.*해/i,
-    script: 'llm-orchestrate.js',
-    args: ['gpt', 'orchestrate', 'You are a software architect. Analyze and review the architecture.'],
-    label: 'gpt-architecture',
-  },
-  {
-    pattern: /(UI|UX).*(리뷰|검토|피드백|개선)|사용자.*경험.*검토|디자인.*리뷰|design.*feedback/i,
-    script: 'llm-orchestrate.js',
-    args: ['gemini', 'orchestrate', 'You are a UI/UX expert. Analyze and provide feedback.'],
-    label: 'gemini-uiux',
-  },
-  {
-    pattern: /디버깅.*해|버그.*찾아|find.*bug|debug.*this.*code/i,
-    script: 'llm-orchestrate.js',
-    args: ['gpt', 'orchestrate', 'You are a debugging expert. Find bugs and suggest fixes.'],
-    label: 'gpt-debug',
-  },
-  {
-    pattern: /코드.*정적.*분석|코드.*분석.*해줘|analyze.*code.*quality/i,
-    script: 'llm-orchestrate.js',
-    args: ['gemini', 'orchestrate', 'You are a code analysis expert. Review and analyze the code.'],
-    label: 'gemini-analysis',
-  },
-  {
-    pattern: /코드.*리뷰|code.*review|PR.*리뷰|리뷰.*해줘.*코드/i,
-    script: 'llm-orchestrate.js',
-    args: ['gpt', 'orchestrate', 'You are a code review expert. Review the code for best practices, security, and performance.'],
-    label: 'gpt-codereview',
-  },
-  {
-    pattern: /추론.*해|reasoning|복잡.*분석|deep.*analysis/i,
-    script: 'llm-orchestrate.js',
-    args: ['gpt', 'orchestrate', 'You are a reasoning expert. Analyze the problem deeply and provide detailed reasoning.'],
-    label: 'gpt-reasoning',
-  },
-
-  // 테스트용
-  {
-    pattern: /^test-gpt/i,
-    script: 'llm-orchestrate.js',
-    args: ['gpt', 'orchestrate', 'You are a helpful assistant. Answer the user\'s question clearly and concisely.'],
-    label: 'test-gpt',
-  },
-  {
-    pattern: /^test-gemini/i,
-    script: 'llm-orchestrate.js',
-    args: ['gemini', 'orchestrate', 'You are a helpful assistant. Answer the user\'s question clearly and concisely.'],
-    label: 'test-gemini',
-  },
-];
-
-// 매칭된 스크립트 실행
-const execPromises = [];
-
-for (const rule of DISPATCH_RULES) {
-  if (rule.label === 'skip') continue;
-
-  // pattern이 null이면 항상 실행, 아니면 매칭 확인
-  if (rule.pattern !== null && !rule.pattern.test(prompt)) continue;
-
-  // echo 규칙: 직접 stdout 출력
-  if (rule.echo) {
-    process.stdout.write(rule.echo + '\n');
-    continue;
-  }
-
-  if (!rule.script) continue;
-
-  const scriptPath = path.join(__dirname, rule.script);
-  const args = rule.args || [];
-
-  execPromises.push(
-    new Promise((resolve) => {
-      execFile('node', [scriptPath, ...args], {
-        timeout: 30000,
-        env: { ...process.env },
-      }, (error, stdout, stderr) => {
-        if (stdout?.trim()) {
-          process.stdout.write(stdout);
-        }
-        resolve();
-      });
-    })
-  );
-}
-
-await Promise.all(execPromises);
-
-// Evolution: Gap detection — log unmatched prompts for skill gap analysis
-const matched = DISPATCH_RULES.some(r =>
-  r.label !== 'skip' && r.label !== 'keyword' &&
-  r.pattern !== null && r.pattern.test(prompt) &&
-  (r.script || r.echo)
-);
-
-if (!matched) {
-  setImmediate(async () => {
-    try {
-      const configPath = path.join(process.env.CLAUDE_PROJECT_DIR || '.', '.claude', 'vibe', 'config.json');
-      let gapEnabled = true;
-      try {
-        const fs = await import('fs');
-        if (fs.existsSync(configPath)) {
-          const config = JSON.parse(fs.readFileSync(configPath, 'utf8'));
-          gapEnabled = config.evolution?.gapDetection !== false && config.evolution?.enabled !== false;
-        }
-      } catch { /* ignore */ }
-
-      if (gapEnabled) {
-        const LIB_BASE = (await import('./utils.js')).getLibBaseUrl();
-        const [memMod, gapMod] = await Promise.all([
-          import(`${LIB_BASE}memory/MemoryStorage.js`),
-          import(`${LIB_BASE}evolution/SkillGapDetector.js`),
-        ]);
-        const storage = new memMod.MemoryStorage(process.env.CLAUDE_PROJECT_DIR || '.');
-        const detector = new gapMod.SkillGapDetector(storage);
-        detector.logMiss(prompt.slice(0, 200));
-        storage.close();
-      }
-    } catch (e) {
-      process.stderr.write(`[Evolution] Gap log error: ${e.message}\n`);
-    }
-  });
-}
+#!/usr/bin/env node
+/**
+ * UserPromptSubmit 디스패처
+ *
+ * UserPromptSubmit은 matcher를 지원하지 않아 모든 hook이 매번 실행됨.
+ * 이 디스패처가 stdin에서 prompt를 읽고, 패턴 매칭 후 해당 스크립트만 실행.
+ *
+ * 이점:
+ * - 외부 LLM 호출(GPT/Gemini)이 패턴 매칭 없이 발동하지 않음
+ * - context window에 불필요한 응답이 주입되지 않음
+ * - 단일 프로세스에서 매칭 후 필요한 스크립트만 fork
+ */
+import { execFile } from 'child_process';
+import { fileURLToPath } from 'url';
+import path from 'path';
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
+
+// 재귀 가드 — 자식 Claude 세션에서 이 hook이 다시 실행되는 것 차단.
+// llm-orchestrate.js의 callClaudeCli가 VIBE_HOOK_DEPTH=1을 주입하므로,
+// 값이 있으면 즉시 종료해 프로세스 폭탄을 막는다.
+if (process.env.VIBE_HOOK_DEPTH) process.exit(0);
+
+// stdin에서 prompt 읽기
+let inputData = '';
+for await (const chunk of process.stdin) {
+  inputData += chunk;
+}
+
+let prompt = '';
+try {
+  const parsed = JSON.parse(inputData);
+  prompt = parsed.prompt || '';
+} catch {
+  process.exit(0);
+}
+
+if (!prompt) process.exit(0);
+
+// 패턴 → 실행할 스크립트 매핑
+// 각 항목: { pattern, script, args, label }
+const DISPATCH_RULES = [
+  // 항상 실행 (경량 스크립트)
+  {
+    pattern: null, // always
+    script: 'keyword-detector.js',
+    args: [prompt],
+    label: 'keyword',
+  },
+
+  // 패턴 매칭이 필요한 스크립트
+  {
+    pattern: /ultrawork|ulw|울트라워크|ralph|ralplan/i,
+    script: null, // keyword-detector가 이미 처리
+    label: 'skip',
+  },
+  // echo 전용 (stdout으로 직접 출력)
+  {
+    pattern: /e2e.*테스트|e2e.*test|playwright|브라우저.*테스트|browser.*test/i,
+    script: null,
+    echo: '[E2E MODE] Use /vibe.utils --e2e for Playwright-based browser testing. Supports visual regression and video recording.',
+    label: 'e2e-echo',
+  },
+
+  // 외부 LLM 호출 (GPT/Gemini) - 패턴 매칭 필수
+  {
+    pattern: /아키텍처.*(검토|리뷰|분석)|architecture.*(review|analyz)|설계.*검토|구조.*분석.*해/i,
+    script: 'llm-orchestrate.js',
+    args: ['gpt', 'orchestrate', 'You are a software architect. Analyze and review the architecture.'],
+    label: 'gpt-architecture',
+  },
+  {
+    pattern: /(UI|UX).*(리뷰|검토|피드백|개선)|사용자.*경험.*검토|디자인.*리뷰|design.*feedback/i,
+    script: 'llm-orchestrate.js',
+    args: ['gemini', 'orchestrate', 'You are a UI/UX expert. Analyze and provide feedback.'],
+    label: 'gemini-uiux',
+  },
+  {
+    pattern: /디버깅.*해|버그.*찾아|find.*bug|debug.*this.*code/i,
+    script: 'llm-orchestrate.js',
+    args: ['gpt', 'orchestrate', 'You are a debugging expert. Find bugs and suggest fixes.'],
+    label: 'gpt-debug',
+  },
+  {
+    pattern: /코드.*정적.*분석|코드.*분석.*해줘|analyze.*code.*quality/i,
+    script: 'llm-orchestrate.js',
+    args: ['gemini', 'orchestrate', 'You are a code analysis expert. Review and analyze the code.'],
+    label: 'gemini-analysis',
+  },
+  {
+    pattern: /코드.*리뷰|code.*review|PR.*리뷰|리뷰.*해줘.*코드/i,
+    script: 'llm-orchestrate.js',
+    args: ['gpt', 'orchestrate', 'You are a code review expert. Review the code for best practices, security, and performance.'],
+    label: 'gpt-codereview',
+  },
+  {
+    pattern: /추론.*해|reasoning|복잡.*분석|deep.*analysis/i,
+    script: 'llm-orchestrate.js',
+    args: ['gpt', 'orchestrate', 'You are a reasoning expert. Analyze the problem deeply and provide detailed reasoning.'],
+    label: 'gpt-reasoning',
+  },
+
+  // 테스트용
+  {
+    pattern: /^test-gpt/i,
+    script: 'llm-orchestrate.js',
+    args: ['gpt', 'orchestrate', 'You are a helpful assistant. Answer the user\'s question clearly and concisely.'],
+    label: 'test-gpt',
+  },
+  {
+    pattern: /^test-gemini/i,
+    script: 'llm-orchestrate.js',
+    args: ['gemini', 'orchestrate', 'You are a helpful assistant. Answer the user\'s question clearly and concisely.'],
+    label: 'test-gemini',
+  },
+];
+
+// 매칭된 스크립트 실행
+const execPromises = [];
+
+for (const rule of DISPATCH_RULES) {
+  if (rule.label === 'skip') continue;
+
+  // pattern이 null이면 항상 실행, 아니면 매칭 확인
+  if (rule.pattern !== null && !rule.pattern.test(prompt)) continue;
+
+  // echo 규칙: 직접 stdout 출력
+  if (rule.echo) {
+    process.stdout.write(rule.echo + '\n');
+    continue;
+  }
+
+  if (!rule.script) continue;
+
+  const scriptPath = path.join(__dirname, rule.script);
+  const args = rule.args || [];
+
+  execPromises.push(
+    new Promise((resolve) => {
+      execFile('node', [scriptPath, ...args], {
+        timeout: 30000,
+        env: { ...process.env },
+      }, (error, stdout, stderr) => {
+        if (stdout?.trim()) {
+          process.stdout.write(stdout);
+        }
+        resolve();
+      });
+    })
+  );
+}
+
+await Promise.all(execPromises);
+
+// Evolution: Gap detection — log unmatched prompts for skill gap analysis
+const matched = DISPATCH_RULES.some(r =>
+  r.label !== 'skip' && r.label !== 'keyword' &&
+  r.pattern !== null && r.pattern.test(prompt) &&
+  (r.script || r.echo)
+);
+
+if (!matched) {
+  setImmediate(async () => {
+    try {
+      const configPath = path.join(process.env.CLAUDE_PROJECT_DIR || '.', '.claude', 'vibe', 'config.json');
+      let gapEnabled = true;
+      try {
+        const fs = await import('fs');
+        if (fs.existsSync(configPath)) {
+          const config = JSON.parse(fs.readFileSync(configPath, 'utf8'));
+          gapEnabled = config.evolution?.gapDetection !== false && config.evolution?.enabled !== false;
+        }
+      } catch { /* ignore */ }
+
+      if (gapEnabled) {
+        const LIB_BASE = (await import('./utils.js')).getLibBaseUrl();
+        const [memMod, gapMod] = await Promise.all([
+          import(`${LIB_BASE}memory/MemoryStorage.js`),
+          import(`${LIB_BASE}evolution/SkillGapDetector.js`),
+        ]);
+        const storage = new memMod.MemoryStorage(process.env.CLAUDE_PROJECT_DIR || '.');
+        const detector = new gapMod.SkillGapDetector(storage);
+        detector.logMiss(prompt.slice(0, 200));
+        storage.close();
+      }
+    } catch (e) {
+      process.stderr.write(`[Evolution] Gap log error: ${e.message}\n`);
+    }
+  });
+}

package/hooks/scripts/sentinel-guard.js

@@ -1,130 +1,130 @@
-#!/usr/bin/env node
-/**
- * Sentinel Guard — PreToolUse hook
- * Protects sentinel files and blocks dangerous operations.
- * Runs before pre-tool-guard.js.
- */
-
-const SENTINEL_PATH_PREFIX = 'src/infra/lib/autonomy/';
-
-const SENTINEL_PATH_RE = /^(\.\/|\.\\)?src[\\/]infra[\\/]lib[\\/]autonomy[\\/]/;
-
-const DANGEROUS_BASH_RE =
-  /\b(rm\s+-rf|kill\s+-9|drop\s+table|truncate|shutdown|reboot|mkfs|dd\s+if=)\b/i;
-
-/**
- * Extract file path from tool input
- */
-function extractFilePath(toolName, input) {
-  if (!input) return null;
-  try {
-    const parsed = typeof input === 'string' ? JSON.parse(input) : input;
-    if (toolName === 'Write' || toolName === 'Edit' || toolName === 'Read') {
-      return parsed.file_path || parsed.filePath || null;
-    }
-  } catch {
-    // Not JSON, try as plain string
-    return typeof input === 'string' ? input : null;
-  }
-  return null;
-}
-
-/**
- * Extract command from Bash tool input
- */
-function extractBashCommand(input) {
-  if (!input) return null;
-  try {
-    const parsed = typeof input === 'string' ? JSON.parse(input) : input;
-    return parsed.command || null;
-  } catch {
-    return typeof input === 'string' ? input : null;
-  }
-}
-
-/**
- * Check if path targets sentinel files
- */
-function isSentinelPath(filePath) {
-  if (!filePath) return false;
-  const normalized = filePath.replace(/\\/g, '/').replace(/^\.\//, '');
-  return normalized.startsWith(SENTINEL_PATH_PREFIX) || SENTINEL_PATH_RE.test(filePath);
-}
-
-/**
- * Main guard logic
- */
-function guard(toolName, toolInput) {
-  // Write/Edit targeting sentinel files → block
-  if (toolName === 'Write' || toolName === 'Edit') {
-    const filePath = extractFilePath(toolName, toolInput);
-    if (isSentinelPath(filePath)) {
-      return {
-        decision: 'block',
-        reason: `Sentinel files are protected. Cannot modify: ${filePath}`,
-      };
-    }
-  }
-
-  // Bash targeting sentinel files or dangerous commands
-  if (toolName === 'Bash') {
-    const command = extractBashCommand(toolInput);
-    if (command && isSentinelPath(command)) {
-      return {
-        decision: 'block',
-        reason: `Sentinel files are protected. Dangerous command targeting sentinel path.`,
-      };
-    }
-    if (command && DANGEROUS_BASH_RE.test(command)) {
-      // Check if command targets sentinel files
-      if (command.includes(SENTINEL_PATH_PREFIX) || command.includes('src/infra/lib/autonomy')) {
-        return {
-          decision: 'block',
-          reason: `Dangerous command targeting sentinel path: ${command}`,
-        };
-      }
-    }
-  }
-
-  // Allow — return undefined for normal flow
-  return undefined;
-}
-
-import fs from 'fs';
-
-/**
- * stdin에서 JSON 페이로드 읽기 (Claude Code 하네스 호환)
- */
-function readStdinSync() {
-  try {
-    if (process.stdin.isTTY) return null;
-    // fd 0을 직접 사용 (Windows는 '/dev/stdin'이 없음)
-    const buf = Buffer.alloc(65536);
-    const bytesRead = fs.readSync(0, buf, 0, buf.length, null);
-    if (bytesRead > 0) {
-      return JSON.parse(buf.toString('utf-8', 0, bytesRead));
-    }
-  } catch { /* 폴백 */ }
-  return null;
-}
-
-// Main execution: stdin JSON 우선, argv 폴백
-const stdinPayload = readStdinSync();
-const toolName = stdinPayload?.tool_name || process.argv[2] || '';
-const toolInput = stdinPayload?.tool_input
-  ? (typeof stdinPayload.tool_input === 'string'
-    ? stdinPayload.tool_input
-    : JSON.stringify(stdinPayload.tool_input))
-  : (process.argv[3] || process.env.TOOL_INPUT || '');
-
-import { logHookDecision } from './utils.js';
-
-const result = guard(toolName, toolInput);
-
-if (result) {
-  logHookDecision('sentinel-guard', toolName, 'block', result.reason);
-  console.log(JSON.stringify(result));
-  process.exit(2); // deny 규약
-}
-
-process.exit(0);
+#!/usr/bin/env node
+/**
+ * Sentinel Guard — PreToolUse hook
+ * Protects sentinel files and blocks dangerous operations.
+ * Runs before pre-tool-guard.js.
+ */
+
+const SENTINEL_PATH_PREFIX = 'src/infra/lib/autonomy/';
+
+const SENTINEL_PATH_RE = /^(\.\/|\.\\)?src[\\/]infra[\\/]lib[\\/]autonomy[\\/]/;
+
+const DANGEROUS_BASH_RE =
+  /\b(rm\s+-rf|kill\s+-9|drop\s+table|truncate|shutdown|reboot|mkfs|dd\s+if=)\b/i;
+
+/**
+ * Extract file path from tool input
+ */
+function extractFilePath(toolName, input) {
+  if (!input) return null;
+  try {
+    const parsed = typeof input === 'string' ? JSON.parse(input) : input;
+    if (toolName === 'Write' || toolName === 'Edit' || toolName === 'Read') {
+      return parsed.file_path || parsed.filePath || null;
+    }
+  } catch {
+    // Not JSON, try as plain string
+    return typeof input === 'string' ? input : null;
+  }
+  return null;
+}
+
+/**
+ * Extract command from Bash tool input
+ */
+function extractBashCommand(input) {
+  if (!input) return null;
+  try {
+    const parsed = typeof input === 'string' ? JSON.parse(input) : input;
+    return parsed.command || null;
+  } catch {
+    return typeof input === 'string' ? input : null;
+  }
+}
+
+/**
+ * Check if path targets sentinel files
+ */
+function isSentinelPath(filePath) {
+  if (!filePath) return false;
+  const normalized = filePath.replace(/\\/g, '/').replace(/^\.\//, '');
+  return normalized.startsWith(SENTINEL_PATH_PREFIX) || SENTINEL_PATH_RE.test(filePath);
+}
+
+/**
+ * Main guard logic
+ */
+function guard(toolName, toolInput) {
+  // Write/Edit targeting sentinel files → block
+  if (toolName === 'Write' || toolName === 'Edit') {
+    const filePath = extractFilePath(toolName, toolInput);
+    if (isSentinelPath(filePath)) {
+      return {
+        decision: 'block',
+        reason: `Sentinel files are protected. Cannot modify: ${filePath}`,
+      };
+    }
+  }
+
+  // Bash targeting sentinel files or dangerous commands
+  if (toolName === 'Bash') {
+    const command = extractBashCommand(toolInput);
+    if (command && isSentinelPath(command)) {
+      return {
+        decision: 'block',
+        reason: `Sentinel files are protected. Dangerous command targeting sentinel path.`,
+      };
+    }
+    if (command && DANGEROUS_BASH_RE.test(command)) {
+      // Check if command targets sentinel files
+      if (command.includes(SENTINEL_PATH_PREFIX) || command.includes('src/infra/lib/autonomy')) {
+        return {
+          decision: 'block',
+          reason: `Dangerous command targeting sentinel path: ${command}`,
+        };
+      }
+    }
+  }
+
+  // Allow — return undefined for normal flow
+  return undefined;
+}
+
+import fs from 'fs';
+
+/**
+ * stdin에서 JSON 페이로드 읽기 (Claude Code 하네스 호환)
+ */
+function readStdinSync() {
+  try {
+    if (process.stdin.isTTY) return null;
+    // fd 0을 직접 사용 (Windows는 '/dev/stdin'이 없음)
+    const buf = Buffer.alloc(65536);
+    const bytesRead = fs.readSync(0, buf, 0, buf.length, null);
+    if (bytesRead > 0) {
+      return JSON.parse(buf.toString('utf-8', 0, bytesRead));
+    }
+  } catch { /* 폴백 */ }
+  return null;
+}
+
+// Main execution: stdin JSON 우선, argv 폴백
+const stdinPayload = readStdinSync();
+const toolName = stdinPayload?.tool_name || process.argv[2] || '';
+const toolInput = stdinPayload?.tool_input
+  ? (typeof stdinPayload.tool_input === 'string'
+    ? stdinPayload.tool_input
+    : JSON.stringify(stdinPayload.tool_input))
+  : (process.argv[3] || process.env.TOOL_INPUT || '');
+
+import { logHookDecision } from './utils.js';
+
+const result = guard(toolName, toolInput);
+
+if (result) {
+  logHookDecision('sentinel-guard', toolName, 'block', result.reason);
+  console.log(JSON.stringify(result));
+  process.exit(2); // deny 규약
+}
+
+process.exit(0);