@vibecheckai/cli 3.8.0 → 3.9.1

package/bin/runners/lib/engine/repo-index.js

@@ -1,514 +1,514 @@
-// bin/runners/lib/engine/repo-index.js
-// Single-pass file indexer with content cache, token prefilter, and framework detection
-
-const fg = require("fast-glob");
-const fs = require("fs");
-const path = require("path");
-const crypto = require("crypto");
-
-// Default ignore patterns (performance-critical)
-const DEFAULT_IGNORE = [
-  "**/node_modules/**",
-  "**/.next/**",
-  "**/dist/**",
-  "**/build/**",
-  "**/.turbo/**",
-  "**/.git/**",
-  "**/.vibecheck/**",
-  "**/.guardrail/**",
-  "**/coverage/**",
-  "**/.nyc_output/**",
-  "**/vendor/**",
-  "**/__pycache__/**",
-  "**/.venv/**",
-  "**/venv/**",
-  "**/env/**",
-  "**/site-packages/**",
-  "**/*.min.js",
-  "**/*.min.css",
-  "**/*.map",
-  "**/*.d.ts",
-];
-
-// Tokens we search for to build a prefilter index
-const PREFILTER_TOKENS = [
-  // Route definitions
-  "fastify", "Fastify", "register", "express", "Express", "app.get", "app.post",
-  "router.get", "router.post", "hono", "Hono", "koa", "Koa",
-  // Client calls
-  "fetch(", "axios", "useSWR", "useQuery",
-  // Python
-  "@app.route", "@router", "FastAPI", "Flask", "Django",
-  // Go
-  "gin.Default", "echo.New", "fiber.New",
-  // Auth
-  "requireAuth", "isAuthenticated", "checkPermission", "authorize",
-  // Env
-  "process.env", "import.meta.env", "os.environ", "getenv",
-];
-
-/**
- * Compute SHA256 hash of content
- * @param {string} content
- * @returns {string}
- */
-function sha256(content) {
-  return crypto.createHash("sha256").update(content).digest("hex");
-}
-
-/**
- * Check if a file exists
- * @param {string} p - Path to check
- * @returns {boolean}
- */
-function fileExists(p) {
-  try {
-    return fs.statSync(p).isFile();
-  } catch {
-    return false;
-  }
-}
-
-/**
- * Check if a directory exists
- * @param {string} p - Path to check
- * @returns {boolean}
- */
-function dirExists(p) {
-  try {
-    return fs.statSync(p).isDirectory();
-  } catch {
-    return false;
-  }
-}
-
-/**
- * RepoIndex - Single-pass file indexer
- * 
- * Usage:
- *   const index = new RepoIndex(repoRoot, { ignorePatterns: [...] });
- *   await index.build();
- *   
- *   // Get all files
- *   const allFiles = index.files;
- *   
- *   // Get files by extension
- *   const tsFiles = index.getByExtension('.ts');
- *   
- *   // Get files containing a token
- *   const fastifyFiles = index.getByToken('fastify');
- *   
- *   // Get file content (cached)
- *   const content = index.getContent(fileAbs);
- *   
- *   // Check signals
- *   if (index.signals.detectedFrameworks.has('fastify')) { ... }
- */
-class RepoIndex {
-  /**
-   * @param {string} repoRoot - Absolute path to repo root
-   * @param {Object} [options]
-   * @param {string[]} [options.ignorePatterns] - Additional ignore patterns
-   * @param {string[]} [options.includePatterns] - File patterns to include (default: all code files)
-   * @param {boolean} [options.buildTokenIndex] - Whether to build token index (default: true)
-   * @param {number} [options.maxFileSize] - Max file size to index in bytes (default: 1MB)
-   */
-  constructor(repoRoot, options = {}) {
-    this.repoRoot = repoRoot;
-    this.ignorePatterns = [...DEFAULT_IGNORE, ...(options.ignorePatterns || [])];
-    this.includePatterns = options.includePatterns || [
-      "**/*.{ts,tsx,js,jsx,mjs,cjs}",
-      "**/*.py",
-      "**/*.go",
-      "**/*.rb",
-      "**/*.json",
-      "**/*.yaml",
-      "**/*.yml",
-    ];
-    this.buildTokenIndex = options.buildTokenIndex !== false;
-    this.maxFileSize = options.maxFileSize || 1024 * 1024; // 1MB
-
-    /** @type {import('./types').FileRecord[]} */
-    this.files = [];
-
-    /** @type {Map<string, string>} abs -> content */
-    this._contentCache = new Map();
-
-    /** @type {Map<string, Set<string>>} token -> Set<abs> */
-    this._tokenIndex = new Map();
-
-    /** @type {Map<string, import('./types').FileRecord[]>} ext -> FileRecord[] */
-    this._byExtension = new Map();
-
-    /** @type {import('./types').RepoSignals} */
-    this.signals = {
-      hasPackageJson: false,
-      hasRequirementsTxt: false,
-      hasPyprojectToml: false,
-      hasGoMod: false,
-      hasGemfile: false,
-      hasNextApp: false,
-      hasNextPages: false,
-      hasOpenAPI: false,
-      hasGraphQL: false,
-      detectedFrameworks: new Set(),
-    };
-
-    this.stats = {
-      totalFiles: 0,
-      totalSize: 0,
-      indexTimeMs: 0,
-      contentCacheHits: 0,
-      contentCacheMisses: 0,
-    };
-
-    this._built = false;
-  }
-
-  /**
-   * Build the index (single glob pass)
-   * @returns {Promise<this>}
-   */
-  async build() {
-    const startTime = Date.now();
-
-    // Single glob pass
-    const filePaths = await fg(this.includePatterns, {
-      cwd: this.repoRoot,
-      absolute: true,
-      ignore: this.ignorePatterns,
-      stats: true,
-      onlyFiles: true,
-    });
-
-    // Process files
-    for (const entry of filePaths) {
-      const abs = typeof entry === "string" ? entry : entry.path;
-      const rel = path.relative(this.repoRoot, abs).replace(/\\/g, "/");
-      const ext = path.extname(abs).toLowerCase();
-
-      let stat;
-      if (typeof entry === "object" && entry.stats) {
-        stat = entry.stats;
-      } else {
-        try {
-          stat = fs.statSync(abs);
-        } catch {
-          continue;
-        }
-      }
-
-      // Skip files that are too large
-      if (stat.size > this.maxFileSize) {
-        continue;
-      }
-
-      const content = this._readFile(abs);
-      if (content === null) continue;
-
-      const hash = sha256(content);
-
-      /** @type {import('./types').FileRecord} */
-      const record = {
-        abs,
-        rel,
-        size: stat.size,
-        mtime: stat.mtimeMs,
-        hash,
-        ext,
-      };
-
-      this.files.push(record);
-      this._contentCache.set(abs, content);
-      this.stats.totalSize += stat.size;
-
-      // Index by extension
-      if (!this._byExtension.has(ext)) {
-        this._byExtension.set(ext, []);
-      }
-      this._byExtension.get(ext).push(record);
-
-      // Build token index
-      if (this.buildTokenIndex) {
-        this._indexTokens(abs, content);
-      }
-    }
-
-    this.stats.totalFiles = this.files.length;
-
-    // Detect signals
-    this._detectSignals();
-
-    this.stats.indexTimeMs = Date.now() - startTime;
-    this._built = true;
-
-    return this;
-  }
-
-  /**
-   * Read file content (with caching)
-   * @param {string} abs - Absolute path
-   * @returns {string|null}
-   */
-  _readFile(abs) {
-    if (this._contentCache.has(abs)) {
-      this.stats.contentCacheHits++;
-      return this._contentCache.get(abs);
-    }
-
-    this.stats.contentCacheMisses++;
-    try {
-      const content = fs.readFileSync(abs, "utf8");
-      return content;
-    } catch {
-      return null;
-    }
-  }
-
-  /**
-   * Index tokens in a file
-   * @param {string} abs
-   * @param {string} content
-   */
-  _indexTokens(abs, content) {
-    for (const token of PREFILTER_TOKENS) {
-      if (content.includes(token)) {
-        if (!this._tokenIndex.has(token)) {
-          this._tokenIndex.set(token, new Set());
-        }
-        this._tokenIndex.get(token).add(abs);
-      }
-    }
-  }
-
-  /**
-   * Detect framework signals from indexed files
-   */
-  _detectSignals() {
-    // Check for manifest files
-    this.signals.hasPackageJson = fileExists(path.join(this.repoRoot, "package.json"));
-    this.signals.hasRequirementsTxt = fileExists(path.join(this.repoRoot, "requirements.txt"));
-    this.signals.hasPyprojectToml = fileExists(path.join(this.repoRoot, "pyproject.toml"));
-    this.signals.hasGoMod = fileExists(path.join(this.repoRoot, "go.mod"));
-    this.signals.hasGemfile = fileExists(path.join(this.repoRoot, "Gemfile"));
-
-    // Check for Next.js directories
-    this.signals.hasNextApp = dirExists(path.join(this.repoRoot, "app")) ||
-      this.files.some(f => f.rel.includes("/app/") && f.rel.includes("route."));
-    this.signals.hasNextPages = dirExists(path.join(this.repoRoot, "pages")) ||
-      this.files.some(f => f.rel.includes("/pages/api/"));
-
-    // Check for OpenAPI/GraphQL
-    this.signals.hasOpenAPI = this.files.some(f =>
-      /openapi\.(json|ya?ml)$/i.test(f.rel) || /swagger\.(json|ya?ml)$/i.test(f.rel)
-    );
-    this.signals.hasGraphQL = this.files.some(f =>
-      /\.graphql$/i.test(f.rel) || /schema\.gql$/i.test(f.rel)
-    );
-
-    // Detect frameworks from package.json
-    if (this.signals.hasPackageJson) {
-      try {
-        const pkg = JSON.parse(fs.readFileSync(path.join(this.repoRoot, "package.json"), "utf8"));
-        const allDeps = { ...pkg.dependencies, ...pkg.devDependencies };
-
-        if (allDeps.fastify) this.signals.detectedFrameworks.add("fastify");
-        if (allDeps.express) this.signals.detectedFrameworks.add("express");
-        if (allDeps.koa) this.signals.detectedFrameworks.add("koa");
-        if (allDeps.hono) this.signals.detectedFrameworks.add("hono");
-        if (allDeps.next) this.signals.detectedFrameworks.add("next");
-        if (allDeps["@nestjs/core"]) this.signals.detectedFrameworks.add("nestjs");
-      } catch {}
-    }
-
-    // Detect Python frameworks
-    if (this.signals.hasRequirementsTxt) {
-      try {
-        const req = fs.readFileSync(path.join(this.repoRoot, "requirements.txt"), "utf8").toLowerCase();
-        if (req.includes("flask")) this.signals.detectedFrameworks.add("flask");
-        if (req.includes("fastapi")) this.signals.detectedFrameworks.add("fastapi");
-        if (req.includes("django")) this.signals.detectedFrameworks.add("django");
-      } catch {}
-    }
-
-    // Detect Go frameworks
-    if (this.signals.hasGoMod) {
-      try {
-        const gomod = fs.readFileSync(path.join(this.repoRoot, "go.mod"), "utf8").toLowerCase();
-        if (gomod.includes("gin-gonic")) this.signals.detectedFrameworks.add("gin");
-        if (gomod.includes("labstack/echo")) this.signals.detectedFrameworks.add("echo");
-        if (gomod.includes("gofiber/fiber")) this.signals.detectedFrameworks.add("fiber");
-        this.signals.detectedFrameworks.add("go");
-      } catch {}
-    }
-
-    // Detect Ruby frameworks
-    if (this.signals.hasGemfile) {
-      try {
-        const gem = fs.readFileSync(path.join(this.repoRoot, "Gemfile"), "utf8").toLowerCase();
-        if (gem.includes("rails")) this.signals.detectedFrameworks.add("rails");
-        if (gem.includes("sinatra")) this.signals.detectedFrameworks.add("sinatra");
-      } catch {}
-    }
-
-    // Add next if we have app/pages directories
-    if (this.signals.hasNextApp || this.signals.hasNextPages) {
-      this.signals.detectedFrameworks.add("next");
-    }
-  }
-
-  /**
-   * Get file content (cached)
-   * @param {string} abs - Absolute path
-   * @returns {string|null}
-   */
-  getContent(abs) {
-    if (this._contentCache.has(abs)) {
-      this.stats.contentCacheHits++;
-      return this._contentCache.get(abs);
-    }
-
-    // Lazy load if not in cache
-    this.stats.contentCacheMisses++;
-    try {
-      const content = fs.readFileSync(abs, "utf8");
-      this._contentCache.set(abs, content);
-      return content;
-    } catch {
-      return null;
-    }
-  }
-
-  /**
-   * Get files by extension
-   * @param {string} ext - Extension with dot (e.g., ".ts")
-   * @returns {import('./types').FileRecord[]}
-   */
-  getByExtension(ext) {
-    return this._byExtension.get(ext.toLowerCase()) || [];
-  }
-
-  /**
-   * Get files containing a token (prefilter)
-   * @param {string} token
-   * @returns {string[]} - Array of absolute paths
-   */
-  getByToken(token) {
-    const set = this._tokenIndex.get(token);
-    return set ? Array.from(set) : [];
-  }
-
-  /**
-   * Get files matching multiple tokens (OR)
-   * @param {string[]} tokens
-   * @returns {string[]} - Array of absolute paths
-   */
-  getByAnyToken(tokens) {
-    const result = new Set();
-    for (const token of tokens) {
-      const set = this._tokenIndex.get(token);
-      if (set) {
-        for (const abs of set) {
-          result.add(abs);
-        }
-      }
-    }
-    return Array.from(result);
-  }
-
-  /**
-   * Get files matching all tokens (AND)
-   * @param {string[]} tokens
-   * @returns {string[]} - Array of absolute paths
-   */
-  getByAllTokens(tokens) {
-    if (tokens.length === 0) return [];
-
-    const sets = tokens.map(t => this._tokenIndex.get(t)).filter(Boolean);
-    if (sets.length === 0) return [];
-    if (sets.length !== tokens.length) return []; // Some tokens not found
-
-    // Intersect all sets
-    let result = new Set(sets[0]);
-    for (let i = 1; i < sets.length; i++) {
-      const next = new Set();
-      for (const abs of result) {
-        if (sets[i].has(abs)) {
-          next.add(abs);
-        }
-      }
-      result = next;
-    }
-
-    return Array.from(result);
-  }
-
-  /**
-   * Get files matching a glob pattern (from already-indexed files)
-   * @param {string} pattern - Glob pattern
-   * @returns {import('./types').FileRecord[]}
-   */
-  getByPattern(pattern) {
-    const picomatch = require("picomatch");
-    const isMatch = picomatch(pattern);
-    return this.files.filter(f => isMatch(f.rel));
-  }
-
-  /**
-   * Check if a framework is detected
-   * @param {string} framework
-   * @returns {boolean}
-   */
-  hasFramework(framework) {
-    return this.signals.detectedFrameworks.has(framework.toLowerCase());
-  }
-
-  /**
-   * Get JS/TS files only
-   * @returns {import('./types').FileRecord[]}
-   */
-  getJsFiles() {
-    return [
-      ...this.getByExtension(".ts"),
-      ...this.getByExtension(".tsx"),
-      ...this.getByExtension(".js"),
-      ...this.getByExtension(".jsx"),
-      ...this.getByExtension(".mjs"),
-      ...this.getByExtension(".cjs"),
-    ];
-  }
-
-  /**
-   * Get Python files only
-   * @returns {import('./types').FileRecord[]}
-   */
-  getPyFiles() {
-    return this.getByExtension(".py");
-  }
-
-  /**
-   * Get Go files only
-   * @returns {import('./types').FileRecord[]}
-   */
-  getGoFiles() {
-    return this.getByExtension(".go");
-  }
-
-  /**
-   * Clear content cache to free memory
-   */
-  clearContentCache() {
-    this._contentCache.clear();
-  }
-
-  /**
-   * Get relative path from absolute
-   * @param {string} abs
-   * @returns {string}
-   */
-  relPath(abs) {
-    return path.relative(this.repoRoot, abs).replace(/\\/g, "/");
-  }
-}
-
-module.exports = { RepoIndex, DEFAULT_IGNORE, PREFILTER_TOKENS };
+// bin/runners/lib/engine/repo-index.js
+// Single-pass file indexer with content cache, token prefilter, and framework detection
+
+const fg = require("fast-glob");
+const fs = require("fs");
+const path = require("path");
+const crypto = require("crypto");
+
+// Default ignore patterns (performance-critical)
+const DEFAULT_IGNORE = [
+  "**/node_modules/**",
+  "**/.next/**",
+  "**/dist/**",
+  "**/build/**",
+  "**/.turbo/**",
+  "**/.git/**",
+  "**/.vibecheck/**",
+  "**/.guardrail/**",
+  "**/coverage/**",
+  "**/.nyc_output/**",
+  "**/vendor/**",
+  "**/__pycache__/**",
+  "**/.venv/**",
+  "**/venv/**",
+  "**/env/**",
+  "**/site-packages/**",
+  "**/*.min.js",
+  "**/*.min.css",
+  "**/*.map",
+  "**/*.d.ts",
+];
+
+// Tokens we search for to build a prefilter index
+const PREFILTER_TOKENS = [
+  // Route definitions
+  "fastify", "Fastify", "register", "express", "Express", "app.get", "app.post",
+  "router.get", "router.post", "hono", "Hono", "koa", "Koa",
+  // Client calls
+  "fetch(", "axios", "useSWR", "useQuery",
+  // Python
+  "@app.route", "@router", "FastAPI", "Flask", "Django",
+  // Go
+  "gin.Default", "echo.New", "fiber.New",
+  // Auth
+  "requireAuth", "isAuthenticated", "checkPermission", "authorize",
+  // Env
+  "process.env", "import.meta.env", "os.environ", "getenv",
+];
+
+/**
+ * Compute SHA256 hash of content
+ * @param {string} content
+ * @returns {string}
+ */
+function sha256(content) {
+  return crypto.createHash("sha256").update(content).digest("hex");
+}
+
+/**
+ * Check if a file exists
+ * @param {string} p - Path to check
+ * @returns {boolean}
+ */
+function fileExists(p) {
+  try {
+    return fs.statSync(p).isFile();
+  } catch {
+    return false;
+  }
+}
+
+/**
+ * Check if a directory exists
+ * @param {string} p - Path to check
+ * @returns {boolean}
+ */
+function dirExists(p) {
+  try {
+    return fs.statSync(p).isDirectory();
+  } catch {
+    return false;
+  }
+}
+
+/**
+ * RepoIndex - Single-pass file indexer
+ * 
+ * Usage:
+ *   const index = new RepoIndex(repoRoot, { ignorePatterns: [...] });
+ *   await index.build();
+ *   
+ *   // Get all files
+ *   const allFiles = index.files;
+ *   
+ *   // Get files by extension
+ *   const tsFiles = index.getByExtension('.ts');
+ *   
+ *   // Get files containing a token
+ *   const fastifyFiles = index.getByToken('fastify');
+ *   
+ *   // Get file content (cached)
+ *   const content = index.getContent(fileAbs);
+ *   
+ *   // Check signals
+ *   if (index.signals.detectedFrameworks.has('fastify')) { ... }
+ */
+class RepoIndex {
+  /**
+   * @param {string} repoRoot - Absolute path to repo root
+   * @param {Object} [options]
+   * @param {string[]} [options.ignorePatterns] - Additional ignore patterns
+   * @param {string[]} [options.includePatterns] - File patterns to include (default: all code files)
+   * @param {boolean} [options.buildTokenIndex] - Whether to build token index (default: true)
+   * @param {number} [options.maxFileSize] - Max file size to index in bytes (default: 1MB)
+   */
+  constructor(repoRoot, options = {}) {
+    this.repoRoot = repoRoot;
+    this.ignorePatterns = [...DEFAULT_IGNORE, ...(options.ignorePatterns || [])];
+    this.includePatterns = options.includePatterns || [
+      "**/*.{ts,tsx,js,jsx,mjs,cjs}",
+      "**/*.py",
+      "**/*.go",
+      "**/*.rb",
+      "**/*.json",
+      "**/*.yaml",
+      "**/*.yml",
+    ];
+    this.buildTokenIndex = options.buildTokenIndex !== false;
+    this.maxFileSize = options.maxFileSize || 1024 * 1024; // 1MB
+
+    /** @type {import('./types').FileRecord[]} */
+    this.files = [];
+
+    /** @type {Map<string, string>} abs -> content */
+    this._contentCache = new Map();
+
+    /** @type {Map<string, Set<string>>} token -> Set<abs> */
+    this._tokenIndex = new Map();
+
+    /** @type {Map<string, import('./types').FileRecord[]>} ext -> FileRecord[] */
+    this._byExtension = new Map();
+
+    /** @type {import('./types').RepoSignals} */
+    this.signals = {
+      hasPackageJson: false,
+      hasRequirementsTxt: false,
+      hasPyprojectToml: false,
+      hasGoMod: false,
+      hasGemfile: false,
+      hasNextApp: false,
+      hasNextPages: false,
+      hasOpenAPI: false,
+      hasGraphQL: false,
+      detectedFrameworks: new Set(),
+    };
+
+    this.stats = {
+      totalFiles: 0,
+      totalSize: 0,
+      indexTimeMs: 0,
+      contentCacheHits: 0,
+      contentCacheMisses: 0,
+    };
+
+    this._built = false;
+  }
+
+  /**
+   * Build the index (single glob pass)
+   * @returns {Promise<this>}
+   */
+  async build() {
+    const startTime = Date.now();
+
+    // Single glob pass
+    const filePaths = await fg(this.includePatterns, {
+      cwd: this.repoRoot,
+      absolute: true,
+      ignore: this.ignorePatterns,
+      stats: true,
+      onlyFiles: true,
+    });
+
+    // Process files
+    for (const entry of filePaths) {
+      const abs = typeof entry === "string" ? entry : entry.path;
+      const rel = path.relative(this.repoRoot, abs).replace(/\\/g, "/");
+      const ext = path.extname(abs).toLowerCase();
+
+      let stat;
+      if (typeof entry === "object" && entry.stats) {
+        stat = entry.stats;
+      } else {
+        try {
+          stat = fs.statSync(abs);
+        } catch {
+          continue;
+        }
+      }
+
+      // Skip files that are too large
+      if (stat.size > this.maxFileSize) {
+        continue;
+      }
+
+      const content = this._readFile(abs);
+      if (content === null) continue;
+
+      const hash = sha256(content);
+
+      /** @type {import('./types').FileRecord} */
+      const record = {
+        abs,
+        rel,
+        size: stat.size,
+        mtime: stat.mtimeMs,
+        hash,
+        ext,
+      };
+
+      this.files.push(record);
+      this._contentCache.set(abs, content);
+      this.stats.totalSize += stat.size;
+
+      // Index by extension
+      if (!this._byExtension.has(ext)) {
+        this._byExtension.set(ext, []);
+      }
+      this._byExtension.get(ext).push(record);
+
+      // Build token index
+      if (this.buildTokenIndex) {
+        this._indexTokens(abs, content);
+      }
+    }
+
+    this.stats.totalFiles = this.files.length;
+
+    // Detect signals
+    this._detectSignals();
+
+    this.stats.indexTimeMs = Date.now() - startTime;
+    this._built = true;
+
+    return this;
+  }
+
+  /**
+   * Read file content (with caching)
+   * @param {string} abs - Absolute path
+   * @returns {string|null}
+   */
+  _readFile(abs) {
+    if (this._contentCache.has(abs)) {
+      this.stats.contentCacheHits++;
+      return this._contentCache.get(abs);
+    }
+
+    this.stats.contentCacheMisses++;
+    try {
+      const content = fs.readFileSync(abs, "utf8");
+      return content;
+    } catch {
+      return null;
+    }
+  }
+
+  /**
+   * Index tokens in a file
+   * @param {string} abs
+   * @param {string} content
+   */
+  _indexTokens(abs, content) {
+    for (const token of PREFILTER_TOKENS) {
+      if (content.includes(token)) {
+        if (!this._tokenIndex.has(token)) {
+          this._tokenIndex.set(token, new Set());
+        }
+        this._tokenIndex.get(token).add(abs);
+      }
+    }
+  }
+
+  /**
+   * Detect framework signals from indexed files
+   */
+  _detectSignals() {
+    // Check for manifest files
+    this.signals.hasPackageJson = fileExists(path.join(this.repoRoot, "package.json"));
+    this.signals.hasRequirementsTxt = fileExists(path.join(this.repoRoot, "requirements.txt"));
+    this.signals.hasPyprojectToml = fileExists(path.join(this.repoRoot, "pyproject.toml"));
+    this.signals.hasGoMod = fileExists(path.join(this.repoRoot, "go.mod"));
+    this.signals.hasGemfile = fileExists(path.join(this.repoRoot, "Gemfile"));
+
+    // Check for Next.js directories
+    this.signals.hasNextApp = dirExists(path.join(this.repoRoot, "app")) ||
+      this.files.some(f => f.rel.includes("/app/") && f.rel.includes("route."));
+    this.signals.hasNextPages = dirExists(path.join(this.repoRoot, "pages")) ||
+      this.files.some(f => f.rel.includes("/pages/api/"));
+
+    // Check for OpenAPI/GraphQL
+    this.signals.hasOpenAPI = this.files.some(f =>
+      /openapi\.(json|ya?ml)$/i.test(f.rel) || /swagger\.(json|ya?ml)$/i.test(f.rel)
+    );
+    this.signals.hasGraphQL = this.files.some(f =>
+      /\.graphql$/i.test(f.rel) || /schema\.gql$/i.test(f.rel)
+    );
+
+    // Detect frameworks from package.json
+    if (this.signals.hasPackageJson) {
+      try {
+        const pkg = JSON.parse(fs.readFileSync(path.join(this.repoRoot, "package.json"), "utf8"));
+        const allDeps = { ...pkg.dependencies, ...pkg.devDependencies };
+
+        if (allDeps.fastify) this.signals.detectedFrameworks.add("fastify");
+        if (allDeps.express) this.signals.detectedFrameworks.add("express");
+        if (allDeps.koa) this.signals.detectedFrameworks.add("koa");
+        if (allDeps.hono) this.signals.detectedFrameworks.add("hono");
+        if (allDeps.next) this.signals.detectedFrameworks.add("next");
+        if (allDeps["@nestjs/core"]) this.signals.detectedFrameworks.add("nestjs");
+      } catch {}
+    }
+
+    // Detect Python frameworks
+    if (this.signals.hasRequirementsTxt) {
+      try {
+        const req = fs.readFileSync(path.join(this.repoRoot, "requirements.txt"), "utf8").toLowerCase();
+        if (req.includes("flask")) this.signals.detectedFrameworks.add("flask");
+        if (req.includes("fastapi")) this.signals.detectedFrameworks.add("fastapi");
+        if (req.includes("django")) this.signals.detectedFrameworks.add("django");
+      } catch {}
+    }
+
+    // Detect Go frameworks
+    if (this.signals.hasGoMod) {
+      try {
+        const gomod = fs.readFileSync(path.join(this.repoRoot, "go.mod"), "utf8").toLowerCase();
+        if (gomod.includes("gin-gonic")) this.signals.detectedFrameworks.add("gin");
+        if (gomod.includes("labstack/echo")) this.signals.detectedFrameworks.add("echo");
+        if (gomod.includes("gofiber/fiber")) this.signals.detectedFrameworks.add("fiber");
+        this.signals.detectedFrameworks.add("go");
+      } catch {}
+    }
+
+    // Detect Ruby frameworks
+    if (this.signals.hasGemfile) {
+      try {
+        const gem = fs.readFileSync(path.join(this.repoRoot, "Gemfile"), "utf8").toLowerCase();
+        if (gem.includes("rails")) this.signals.detectedFrameworks.add("rails");
+        if (gem.includes("sinatra")) this.signals.detectedFrameworks.add("sinatra");
+      } catch {}
+    }
+
+    // Add next if we have app/pages directories
+    if (this.signals.hasNextApp || this.signals.hasNextPages) {
+      this.signals.detectedFrameworks.add("next");
+    }
+  }
+
+  /**
+   * Get file content (cached)
+   * @param {string} abs - Absolute path
+   * @returns {string|null}
+   */
+  getContent(abs) {
+    if (this._contentCache.has(abs)) {
+      this.stats.contentCacheHits++;
+      return this._contentCache.get(abs);
+    }
+
+    // Lazy load if not in cache
+    this.stats.contentCacheMisses++;
+    try {
+      const content = fs.readFileSync(abs, "utf8");
+      this._contentCache.set(abs, content);
+      return content;
+    } catch {
+      return null;
+    }
+  }
+
+  /**
+   * Get files by extension
+   * @param {string} ext - Extension with dot (e.g., ".ts")
+   * @returns {import('./types').FileRecord[]}
+   */
+  getByExtension(ext) {
+    return this._byExtension.get(ext.toLowerCase()) || [];
+  }
+
+  /**
+   * Get files containing a token (prefilter)
+   * @param {string} token
+   * @returns {string[]} - Array of absolute paths
+   */
+  getByToken(token) {
+    const set = this._tokenIndex.get(token);
+    return set ? Array.from(set) : [];
+  }
+
+  /**
+   * Get files matching multiple tokens (OR)
+   * @param {string[]} tokens
+   * @returns {string[]} - Array of absolute paths
+   */
+  getByAnyToken(tokens) {
+    const result = new Set();
+    for (const token of tokens) {
+      const set = this._tokenIndex.get(token);
+      if (set) {
+        for (const abs of set) {
+          result.add(abs);
+        }
+      }
+    }
+    return Array.from(result);
+  }
+
+  /**
+   * Get files matching all tokens (AND)
+   * @param {string[]} tokens
+   * @returns {string[]} - Array of absolute paths
+   */
+  getByAllTokens(tokens) {
+    if (tokens.length === 0) return [];
+
+    const sets = tokens.map(t => this._tokenIndex.get(t)).filter(Boolean);
+    if (sets.length === 0) return [];
+    if (sets.length !== tokens.length) return []; // Some tokens not found
+
+    // Intersect all sets
+    let result = new Set(sets[0]);
+    for (let i = 1; i < sets.length; i++) {
+      const next = new Set();
+      for (const abs of result) {
+        if (sets[i].has(abs)) {
+          next.add(abs);
+        }
+      }
+      result = next;
+    }
+
+    return Array.from(result);
+  }
+
+  /**
+   * Get files matching a glob pattern (from already-indexed files)
+   * @param {string} pattern - Glob pattern
+   * @returns {import('./types').FileRecord[]}
+   */
+  getByPattern(pattern) {
+    const picomatch = require("picomatch");
+    const isMatch = picomatch(pattern);
+    return this.files.filter(f => isMatch(f.rel));
+  }
+
+  /**
+   * Check if a framework is detected
+   * @param {string} framework
+   * @returns {boolean}
+   */
+  hasFramework(framework) {
+    return this.signals.detectedFrameworks.has(framework.toLowerCase());
+  }
+
+  /**
+   * Get JS/TS files only
+   * @returns {import('./types').FileRecord[]}
+   */
+  getJsFiles() {
+    return [
+      ...this.getByExtension(".ts"),
+      ...this.getByExtension(".tsx"),
+      ...this.getByExtension(".js"),
+      ...this.getByExtension(".jsx"),
+      ...this.getByExtension(".mjs"),
+      ...this.getByExtension(".cjs"),
+    ];
+  }
+
+  /**
+   * Get Python files only
+   * @returns {import('./types').FileRecord[]}
+   */
+  getPyFiles() {
+    return this.getByExtension(".py");
+  }
+
+  /**
+   * Get Go files only
+   * @returns {import('./types').FileRecord[]}
+   */
+  getGoFiles() {
+    return this.getByExtension(".go");
+  }
+
+  /**
+   * Clear content cache to free memory
+   */
+  clearContentCache() {
+    this._contentCache.clear();
+  }
+
+  /**
+   * Get relative path from absolute
+   * @param {string} abs
+   * @returns {string}
+   */
+  relPath(abs) {
+    return path.relative(this.repoRoot, abs).replace(/\\/g, "/");
+  }
+}
+
+module.exports = { RepoIndex, DEFAULT_IGNORE, PREFILTER_TOKENS };