@contentrain/mcp 1.1.2 → 1.2.0

package/dist/scanner-CQOxARUU.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"scanner-CQOxARUU.mjs","names":[],"sources":["../src/core/ast-scanner/pre-filter.ts","../src/core/ast-scanner/index.ts","../src/core/scanner.ts"],"sourcesContent":["import type { ExtractedString } from './types.js'\n\n// ─── Pre-filter Result ───\n\nexport interface PreFilterResult {\n  /** Strings that passed the pre-filter (candidates for agent) */\n  candidates: ExtractedString[]\n  /** Total number of strings removed by shouldSkip */\n  skipped: number\n  /** Total number of strings removed by low content score */\n  lowConfidence: number\n  /** Breakdown: skip reason → count */\n  skipReasons: Record<string, number>\n}\n\n// ─── Value-based regexes ───\n\nconst PURE_NUMBER_RE = /^-?\\d+(\\.\\d+)?$/\nconst HEX_COLOR_RE = /^#[0-9a-f]{3,8}$/i\nconst FILE_EXT_RE = /\\.(png|jpg|jpeg|gif|svg|webp|ico|css|scss|less|js|ts|tsx|jsx|json|md|html|xml|yaml|yml|woff|woff2|ttf|eot|mp4|webm|mp3|wav|pdf)$/i\nconst SVG_PATH_DATA_RE = /^[Mm][\\d\\s.,LHVCSQTAZlhvcsqtazmMzZ-]+$/\nconst SVG_VIEWBOX_RE = /^\\d+(\\.\\d+)?\\s+\\d+(\\.\\d+)?\\s+\\d+(\\.\\d+)?\\s+\\d+(\\.\\d+)?$/\nconst I18N_KEY_RE = /^[a-z][a-z0-9_]*(\\.[a-z][a-z0-9_]*)+$/\nconst TECHNICAL_IDENTIFIER_RE = /^[a-z][a-z0-9_-]*$/\nconst ERROR_CODE_RE = /^[A-Z][A-Z0-9_]+$/\nconst PLACEHOLDER_RE = /^\\{\\d+\\}$|^\\.{2,}$/\nconst CAMEL_CASE_RE = /^[a-z]+[A-Z]/\n\n// ─── URL / path detection (consolidated from legacy isNonContent) ───\n\nfunction isURLLike(str: string): boolean {\n  if (/^(https?|ftp|file|mailto|data):/.test(str)) return true\n  if (/^(\\.\\.?\\/|\\/|[A-Za-z]:\\\\)/.test(str)) return true\n  if (/^['\"]?[@a-z][\\w-]*/.test(str.toLowerCase()) && !str.includes(' ') && (str.includes('/') || str.includes('.'))) {\n    return true\n  }\n  return false\n}\n\n// ─── CSS / Tailwind detection ───\n\nconst TAILWIND_SEGMENT_RE = /^(?:bg-|text-|border-|flex|grid|p-|px-|py-|pt-|pb-|pl-|pr-|m-|mx-|my-|mt-|mb-|ml-|mr-|rounded|shadow|w-|h-|min-|max-|gap-|space-|items-|justify-|self-|overflow-|z-|opacity-|transition|duration-|ease-|animate-|font-|leading-|tracking-|decoration-|underline|line-through|uppercase|lowercase|capitalize|truncate|whitespace-|break-|sr-only|not-sr-only|hover:|focus:|active:|disabled:|dark:|sm:|md:|lg:|xl:|2xl:|group-|peer-|ring-|outline-|divide-|table-|col-|row-|aspect-|object-|inset-|top-|right-|bottom-|left-|translate-|rotate-|scale-|skew-|origin-|cursor-|select-|resize-|fill-|stroke-|block|inline|absolute|relative|fixed|sticky|static|float-|clear-|isolate|visible|invisible|grow|shrink|basis-|order-|place-)/\n\nfunction isCssClassList(value: string): boolean {\n  const segments = value.trim().split(/\\s+/)\n  if (segments.length < 2) return false\n  let matched = 0\n  for (const seg of segments) {\n    if (TAILWIND_SEGMENT_RE.test(seg)) matched++\n  }\n  return matched / segments.length >= 0.5\n}\n\nfunction isSingleCssUtility(value: string): boolean {\n  const trimmed = value.trim()\n  if (trimmed.includes(' ')) return false\n  return TAILWIND_SEGMENT_RE.test(trimmed)\n}\n\n// ─── SVG technical attributes ───\n\nconst SVG_TECHNICAL_ATTRIBUTES = new Set([\n  'd', 'viewBox', 'points', 'transform', 'pathLength',\n  'xmlns', 'preserveAspectRatio',\n  'stroke-linecap', 'stroke-linejoin', 'stroke-width',\n  'stroke-dasharray', 'stroke-dashoffset', 'stroke-miterlimit',\n  'fill-rule', 'clip-rule',\n])\n\nconst SVG_GRAPHIC_ELEMENTS = new Set([\n  'svg', 'path', 'circle', 'rect', 'line', 'polyline', 'polygon',\n  'ellipse', 'g', 'defs', 'use', 'symbol', 'clipPath', 'mask',\n  'pattern', 'linearGradient', 'radialGradient', 'stop',\n  'marker', 'animate', 'animateTransform', 'image',\n])\n\n// ─── Known function names ───\n\nconst I18N_FUNCTIONS = new Set([\n  't', '$t', 'i18n', 'translate', 'formatMessage', 'msg',\n])\n\nconst EMIT_FUNCTIONS = new Set([\n  'emit', '$emit',\n])\n\n// ─── Translatable attribute whitelist (i18next-cli compatible + extended) ───\n\nconst TRANSLATABLE_ATTRIBUTES = new Set([\n  // Standard HTML content attributes\n  'title', 'alt', 'placeholder', 'label', 'summary', 'caption',\n  'abbr', 'accesskey', 'content', 'description',\n  // ARIA content\n  'aria-label', 'aria-description', 'aria-placeholder',\n  'aria-roledescription', 'aria-valuetext',\n  // React Native accessibility (equivalent to aria-label)\n  'accessibilityLabel', 'accessibilityHint', 'accessibilityValue',\n  // Common component content props\n  'heading', 'subheading', 'message', 'hint', 'tooltip',\n  'helper-text', 'error-message', 'success-message',\n  'confirm-text', 'cancel-text', 'empty-text', 'loading-text',\n  'no-data-text', 'no-results-text',\n])\n\n// ─── Translatable object property whitelist (i18next-cli compatible) ───\n\nconst TRANSLATABLE_PROPERTIES = new Set([\n  'label', 'title', 'description', 'text', 'message', 'placeholder',\n  'caption', 'summary', 'heading', 'subheading', 'subtitle', 'tooltip',\n  'hint', 'helpText', 'errorMessage', 'successMessage', 'name',\n])\n\n// ─── shouldSkip: Binary non-content detection ───\n\n/**\n * Determines if a string is definitely NOT user-visible content.\n * Returns skip reason if it should be filtered, null if it should proceed to scoring.\n *\n * Conservative for template_text/jsx_text (tag-between text is almost always content).\n * Aggressive for everything else (technical tokens, config values, framework artifacts).\n */\nexport function shouldSkip(str: ExtractedString): string | null {\n  // ── Context-based rules (AST-determined, 100% accurate) ──\n\n  if (str.context === 'import_path') return 'import_path'\n  if (str.context === 'type_annotation') return 'type_annotation'\n  if (str.context === 'css_class') return 'css_class'\n  if (str.context === 'css_utility_call') return 'css_utility_call'\n  if (str.context === 'console_call') return 'console_call'\n  if (str.context === 'test_assertion') return 'test_assertion'\n  if (str.context === 'switch_case') return 'switch_case'\n\n  const v = str.value\n\n  // ── Value-based rules (structural patterns) ──\n\n  if (v.length <= 1) return 'single_char'\n  if (/^\\s+$/.test(v)) return 'whitespace'\n  if (PURE_NUMBER_RE.test(v)) return 'pure_number'\n  if (HEX_COLOR_RE.test(v)) return 'hex_color'\n  if (FILE_EXT_RE.test(v)) return 'file_extension'\n  if (v.startsWith('--')) return 'cli_flag'\n\n  // ── i18n key paths (checked before URL — both contain dots, but i18n keys are more specific) ──\n\n  if (I18N_KEY_RE.test(v)) return 'i18n_key'\n\n  // ── URL/path patterns ──\n\n  if (isURLLike(v)) return 'url_path'\n\n  // ── CSS patterns ──\n\n  if (isCssClassList(v)) return 'css_class_list'\n  if (isSingleCssUtility(v)) return 'css_utility_token'\n\n  // ── SVG patterns ──\n\n  if (v.length > 3 && SVG_PATH_DATA_RE.test(v)) return 'svg_path_data'\n  if (SVG_VIEWBOX_RE.test(v)) return 'svg_viewbox'\n  if (str.parentProperty !== undefined && SVG_TECHNICAL_ATTRIBUTES.has(str.parentProperty)) return 'svg_technical_attr'\n  if (str.context === 'template_attribute' && SVG_GRAPHIC_ELEMENTS.has(str.parent)) return 'svg_element_attr'\n\n  // ── Framework event patterns ──\n\n  if (v.startsWith('update:')) return 'vue_emit_event'\n\n  // ── Placeholder / interpolation ──\n\n  if (PLACEHOLDER_RE.test(v)) return 'placeholder'\n\n  // ── Known function argument detection ──\n\n  if (str.context === 'function_argument' && I18N_FUNCTIONS.has(str.parent)) {\n    // i18n function args: filter lowercase identifiers (namespace/key), keep sentences\n    if (/^[a-z][a-z0-9_.-]*$/.test(v)) return 'i18n_function_arg'\n  }\n\n  if (str.context === 'function_argument' && EMIT_FUNCTIONS.has(str.parent)) {\n    return 'emit_event_arg'\n  }\n\n  // ── CRITICAL: Technical identifier detection (i18next-cli proven pattern) ──\n  // Single lowercase ASCII word/kebab-case/snake_case < 30 chars → technical token\n  // EXEMPT: template_text and jsx_text (tag-between text IS content, even lowercase)\n\n  if (str.context !== 'template_text' && str.context !== 'jsx_text') {\n    if (TECHNICAL_IDENTIFIER_RE.test(v) && v.length < 30) {\n      return 'technical_identifier'\n    }\n  }\n\n  // ── Error codes (SCREAMING_SNAKE_CASE with underscores) ──\n\n  if (ERROR_CODE_RE.test(v) && v.includes('_') && v.length > 3) {\n    return 'error_code'\n  }\n\n  return null\n}\n\n// ─── calculateContentScore: 0-1 confidence scoring ───\n\n/**\n * Calculates a content confidence score (0-1) for a string that passed shouldSkip.\n * Uses AST context metadata (our advantage over offset-based tools) combined with\n * value-based signals proven by i18next-cli.\n *\n * Base score: 0.5. Boosted/penalized by context and value characteristics.\n */\nexport function calculateContentScore(str: ExtractedString): number {\n  let score = 0.5\n\n  // ── Context signals (AST metadata advantage) ──\n\n  // Template/JSX text = almost certainly user-visible content\n  if (str.context === 'template_text' || str.context === 'jsx_text') {\n    score += 0.3\n  }\n\n  // Content-bearing attribute (title, alt, placeholder, aria-label, etc.)\n  if (str.context === 'template_attribute' || str.context === 'jsx_attribute') {\n    if (str.parentProperty && TRANSLATABLE_ATTRIBUTES.has(str.parentProperty)) {\n      score += 0.2\n    } else {\n      score -= 0.2  // Unknown/technical attribute\n    }\n  }\n\n  // Content-bearing object property (message, label, description, etc.)\n  if (str.context === 'object_property') {\n    if (str.parentProperty && TRANSLATABLE_PROPERTIES.has(str.parentProperty)) {\n      score += 0.25\n    }\n  }\n\n  // ── Value signals (i18next-cli proven heuristics) ──\n\n  // Multi-word strings are more likely content\n  const wordCount = str.value.split(/\\s+/).length\n  if (wordCount >= 3) score += 0.2\n  else if (wordCount === 2) score += 0.1\n\n  // Terminal punctuation suggests a sentence\n  if (/[.!?:;]$/.test(str.value)) score += 0.1\n\n  // Non-ASCII characters (Turkish, Chinese, Arabic, etc.) → almost certainly content\n  if (/[\\u0080-\\uFFFF]/.test(str.value)) score += 0.15\n\n  // Capitalized first letter with lowercase body (Dashboard, Kaydet, Settings)\n  if (/^[A-Z]/.test(str.value) && /[a-z]/.test(str.value)) score += 0.1\n\n  // camelCase → probably a technical identifier\n  if (CAMEL_CASE_RE.test(str.value)) score -= 0.3\n\n  // Contains slash without spaces → path-like\n  if (str.value.includes('/') && !str.value.includes(' ')) score -= 0.2\n\n  return Math.max(0, Math.min(1, score))\n}\n\n// ─── Public API ───\n\n/**\n * Two-phase pre-filter:\n * 1. shouldSkip(): Binary removal of definite non-content\n * 2. calculateContentScore(): 0-1 confidence scoring for ambiguous strings\n *\n * Returns candidates that passed both phases, with content scores attached.\n */\nexport function applyPreFilter(\n  strings: ExtractedString[],\n  minScore: number = 0.4,\n): PreFilterResult {\n  const candidates: ExtractedString[] = []\n  const skipReasons: Record<string, number> = {}\n  let skipped = 0\n  let lowConfidence = 0\n\n  for (const str of strings) {\n    // Phase 1: Binary skip\n    const skipReason = shouldSkip(str)\n    if (skipReason) {\n      skipped++\n      skipReasons[skipReason] = (skipReasons[skipReason] ?? 0) + 1\n      continue\n    }\n\n    // Phase 2: Content scoring\n    const contentScore = calculateContentScore(str)\n    if (contentScore < minScore) {\n      lowConfidence++\n      skipReasons['low_confidence'] = (skipReasons['low_confidence'] ?? 0) + 1\n      continue\n    }\n\n    // Attach score to the extraction for downstream use\n    ;(str as ExtractedString & { contentScore: number }).contentScore = contentScore\n    candidates.push(str)\n  }\n\n  return { candidates, skipped, lowConfidence, skipReasons }\n}\n","import type { ExtractedString } from './types.js'\nimport { applyPreFilter } from './pre-filter.js'\nimport { parseTsx } from './tsx-parser.js'\n\n// ─── Extension sets ───\n\nconst TSX_EXTENSIONS = new Set(['.tsx', '.jsx', '.ts', '.js', '.mjs'])\nconst VUE_EXTENSIONS = new Set(['.vue'])\nconst SVELTE_EXTENSIONS = new Set(['.svelte'])\nconst ASTRO_EXTENSIONS = new Set(['.astro'])\n\n// ─── Lazy-loaded parsers ───\n\n/**\n * Lazily import vue-parser. Returns undefined if not available\n * (@vue/compiler-sfc is an optional dependency).\n */\nasync function loadVueParser(): Promise<((content: string, fileName: string) => Promise<ExtractedString[]> | ExtractedString[]) | undefined> {\n  try {\n    const mod = await import('./vue-parser.js')\n    return mod.parseVue\n  } catch {\n    return undefined\n  }\n}\n\n/**\n * Lazily import svelte-parser. Returns undefined if not available\n * (svelte is an optional dependency).\n */\nasync function loadSvelteParser(): Promise<((content: string, fileName: string) => Promise<ExtractedString[]>) | undefined> {\n  try {\n    const mod = await import('./svelte-parser.js')\n    return mod.parseSvelte\n  } catch {\n    return undefined\n  }\n}\n\n/**\n * Lazily import astro-parser. Returns undefined if not available\n * (@astrojs/compiler is an optional dependency).\n */\nasync function loadAstroParser(): Promise<((content: string, fileName: string) => Promise<ExtractedString[]>) | undefined> {\n  try {\n    const mod = await import('./astro-parser.js')\n    return mod.parseAstro\n  } catch {\n    return undefined\n  }\n}\n\n// ─── Regex fallback for unknown extensions ───\n\nconst SURROUNDING_MAX = 120\n\n/**\n * Minimal regex-based extractor for file types without AST parsers.\n * Extracts quoted strings and tag text — conservative, low accuracy.\n * Used as fallback when dedicated parsers are unavailable.\n */\nfunction extractWithRegex(content: string, _filePath: string): ExtractedString[] {\n  const lines = content.split('\\n')\n  const results: ExtractedString[] = []\n\n  // Simple quoted string extraction\n  const stringRe = /(['\"`])(?:(?!\\1|\\\\).|\\\\.)*?\\1/g\n\n  for (let i = 0; i < lines.length; i++) {\n    const line = lines[i]!\n    const trimmed = line.trim()\n\n    // Skip comments, imports\n    if (trimmed.startsWith('//') || trimmed.startsWith('/*') || trimmed.startsWith('*')) continue\n    if (/^\\s*(import|export)\\s/.test(line)) continue\n\n    let match: RegExpExecArray | null\n    stringRe.lastIndex = 0\n\n    while ((match = stringRe.exec(line)) !== null) {\n      // Remove surrounding quotes\n      const raw = match[0]\n      const value = raw.slice(1, -1)\n      if (value.length === 0) continue\n\n      const surrounding = buildSurrounding(lines, i)\n\n      results.push({\n        value,\n        line: i + 1,\n        column: match.index + 1,\n        context: 'other',\n        scope: 'script',\n        parent: '',\n        surrounding,\n      })\n    }\n\n    // Tag text extraction: >text<\n    const tagTextRe = />([^<>{]+)</g\n    let tagMatch: RegExpExecArray | null\n    tagTextRe.lastIndex = 0\n\n    while ((tagMatch = tagTextRe.exec(line)) !== null) {\n      const text = tagMatch[1]!.trim()\n      if (text.length === 0) continue\n      if (/^[\\s\\W]*$/.test(text) && !/[a-zA-Z]/.test(text)) continue\n\n      results.push({\n        value: text,\n        line: i + 1,\n        column: tagMatch.index + 1,\n        context: 'template_text',\n        scope: 'template',\n        parent: '',\n        surrounding: buildSurrounding(lines, i),\n      })\n    }\n  }\n\n  return results\n}\n\nfunction buildSurrounding(lines: string[], lineIdx: number): string {\n  const start = Math.max(0, lineIdx - 1)\n  const end = Math.min(lines.length - 1, lineIdx + 1)\n\n  const parts: string[] = []\n  for (let i = start; i <= end; i++) {\n    const line = lines[i]\n    if (line !== undefined) {\n      parts.push(line)\n    }\n  }\n\n  const joined = parts.join('\\n')\n  if (joined.length > SURROUNDING_MAX) {\n    return joined.slice(0, SURROUNDING_MAX)\n  }\n  return joined\n}\n\n// ─── Public API ───\n\n/**\n * Extract strings from a source file with structural context.\n *\n * Routes to the correct parser based on file extension:\n * - .tsx/.jsx/.ts/.js/.mjs -> tsx-parser (AST-based)\n * - .vue -> vue-parser (lazy-loaded, AST-based)\n * - .svelte -> svelte-parser (lazy-loaded, AST-based)\n * - .astro -> astro-parser (lazy-loaded, AST-based)\n * - unknown -> empty array\n *\n * Falls back to regex extraction when the dedicated parser's\n * optional dependency is not installed.\n *\n * Applies structural pre-filter to remove 100% non-content strings.\n * Returns only candidates that should be sent to the agent.\n */\nexport async function extractStrings(\n  filePath: string,\n  content: string,\n  ext: string,\n): Promise<ExtractedString[]> {\n  const normalizedExt = ext.startsWith('.') ? ext : `.${ext}`\n  let rawStrings: ExtractedString[]\n\n  if (TSX_EXTENSIONS.has(normalizedExt)) {\n    rawStrings = parseTsx(content, filePath)\n  } else if (VUE_EXTENSIONS.has(normalizedExt)) {\n    const parseVue = await loadVueParser()\n    if (parseVue) {\n      rawStrings = await parseVue(content, filePath)\n    } else {\n      rawStrings = extractWithRegex(content, filePath)\n    }\n  } else if (SVELTE_EXTENSIONS.has(normalizedExt)) {\n    const parseSvelte = await loadSvelteParser()\n    if (parseSvelte) {\n      try {\n        rawStrings = await parseSvelte(content, filePath)\n      } catch {\n        // svelte/compiler not installed — fall back to regex\n        rawStrings = extractWithRegex(content, filePath)\n      }\n    } else {\n      rawStrings = extractWithRegex(content, filePath)\n    }\n  } else if (ASTRO_EXTENSIONS.has(normalizedExt)) {\n    const parseAstro = await loadAstroParser()\n    if (parseAstro) {\n      try {\n        rawStrings = await parseAstro(content, filePath)\n      } catch {\n        // @astrojs/compiler not installed — fall back to regex\n        rawStrings = extractWithRegex(content, filePath)\n      }\n    } else {\n      rawStrings = extractWithRegex(content, filePath)\n    }\n  } else {\n    return []\n  }\n\n  // Apply structural pre-filter\n  const { candidates } = applyPreFilter(rawStrings)\n  return candidates\n}\n\n// Re-export types for convenience\nexport type { ExtractedString, StructuralContext, PreFilterRule } from './types.js'\nexport { applyPreFilter, shouldSkip, calculateContentScore } from './pre-filter.js'\nexport type { PreFilterResult } from './pre-filter.js'\n","import type {\n  StringContext,\n  ScanCandidate,\n  DuplicateGroup,\n  ScanCandidatesResult,\n  ScanSummaryResult,\n} from '@contentrain/types'\nimport { join, extname } from 'node:path'\nimport { readText } from '../util/fs.js'\nimport {\n  autoDetectSourceDirs,\n  discoverFiles,\n} from './scan-config.js'\nimport { extractStrings } from './ast-scanner/index.js'\nimport { calculateContentScore } from './ast-scanner/pre-filter.js'\nimport type { StructuralContext } from './ast-scanner/types.js'\n\n// ─── Options ───\n\nexport interface ScanOptions {\n  paths?: string[]\n  include?: string[]\n  exclude?: string[]\n  limit?: number\n  offset?: number\n  min_length?: number\n  max_length?: number\n  /** Minimum content confidence score (0-1). Default: 0.4 */\n  min_score?: number\n}\n\n// ─── Constants ───\n\nconst DEFAULT_LIMIT = 50\nconst DEFAULT_OFFSET = 0\nconst DEFAULT_MIN_LENGTH = 2\nconst DEFAULT_MAX_LENGTH = 500\nconst DEFAULT_MIN_SCORE = 0.4\nconst SUMMARY_SAMPLE_SIZE = 10\n\n// ─── Context mapping: StructuralContext → StringContext ───\n\nconst CONTEXT_MAP: Record<StructuralContext, StringContext> = {\n  'template_text': 'template_text',\n  'template_attribute': 'template_attribute',\n  'jsx_text': 'jsx_text',\n  'jsx_attribute': 'jsx_attribute',\n  'variable_assignment': 'variable_assignment',\n  'object_property': 'object_value',\n  'function_argument': 'function_argument',\n  'array_element': 'other',\n  'enum_value': 'other',\n  'template_literal': 'other',\n  'switch_case': 'other',\n  'other': 'other',\n  // Pre-filtered contexts should not reach here, but map them just in case\n  'import_path': 'other',\n  'type_annotation': 'other',\n  'css_class': 'other',\n  'css_utility_call': 'other',\n  'console_call': 'other',\n  'test_assertion': 'other',\n}\n\n// ─── Main: scanCandidates ───\n\nexport async function scanCandidates(\n  projectRoot: string,\n  options?: ScanOptions,\n): Promise<ScanCandidatesResult> {\n  const limit = options?.limit ?? DEFAULT_LIMIT\n  const offset = options?.offset ?? DEFAULT_OFFSET\n  const minLength = options?.min_length ?? DEFAULT_MIN_LENGTH\n  const maxLength = options?.max_length ?? DEFAULT_MAX_LENGTH\n  const minScore = options?.min_score ?? DEFAULT_MIN_SCORE\n\n  const scanDirs = options?.paths ?? await autoDetectSourceDirs(projectRoot)\n  const files = await discoverFiles(projectRoot, {\n    paths: scanDirs,\n    include: options?.include,\n    exclude: options?.exclude,\n  })\n\n  // ─── Phase 1: Extract strings from all files (pre-filter applied inside extractStrings) ───\n\n  const filePromises = files.map(async (relPath) => {\n    const filePath = join(projectRoot, relPath)\n    const content = await readText(filePath)\n    if (!content) return { relPath, extractions: [] }\n\n    const ext = extname(filePath)\n    const extractions = await extractStrings(filePath, content, ext)\n    return { relPath, extractions }\n  })\n\n  const fileResults = await Promise.all(filePromises)\n\n  // ─── Phase 2: Score + length filter + deduplicate ───\n\n  let rawStringsFound = 0\n  let skippedCount = 0\n  let lowConfidenceCount = 0\n  const skipReasons: Record<string, number> = {}\n\n  // Deduplication map: value → first candidate + all occurrences\n  const uniqueMap = new Map<string, {\n    candidate: ScanCandidate\n    maxScore: number\n  }>()\n\n  // Also track duplicates for backward compatibility\n  const dupeMap = new Map<string, Array<{ file: string; line: number }>>()\n\n  for (const { relPath, extractions } of fileResults) {\n    rawStringsFound += extractions.length\n\n    for (const extraction of extractions) {\n      // Length filter\n      if (extraction.value.length < minLength || extraction.value.length > maxLength) {\n        skippedCount++\n        skipReasons['length_filter'] = (skipReasons['length_filter'] ?? 0) + 1\n        continue\n      }\n\n      // shouldSkip was already applied inside extractStrings (via applyPreFilter).\n      // But applyPreFilter uses default minScore. Here we apply the user-configured minScore\n      // on the contentScore that was attached during pre-filtering.\n\n      // Get the contentScore attached by applyPreFilter\n      const contentScore: number = (extraction as unknown as { contentScore?: number }).contentScore ?? calculateContentScore(extraction)\n\n      if (contentScore < minScore) {\n        lowConfidenceCount++\n        skipReasons['low_confidence'] = (skipReasons['low_confidence'] ?? 0) + 1\n        continue\n      }\n\n      const mappedContext = CONTEXT_MAP[extraction.context]\n      const loc = { file: relPath, line: extraction.line }\n\n      // Track all occurrences for duplicates section\n      if (!dupeMap.has(extraction.value)) {\n        dupeMap.set(extraction.value, [])\n      }\n      dupeMap.get(extraction.value)!.push(loc)\n\n      // Deduplication: keep first occurrence, accumulate locations\n      if (!uniqueMap.has(extraction.value)) {\n        uniqueMap.set(extraction.value, {\n          candidate: {\n            file: relPath,\n            line: extraction.line,\n            column: extraction.column,\n            value: extraction.value,\n            context: mappedContext,\n            surrounding: extraction.surrounding,\n            contentScore,\n            occurrences: [loc],\n          },\n          maxScore: contentScore,\n        })\n      } else {\n        const entry = uniqueMap.get(extraction.value)!\n        entry.candidate.occurrences.push(loc)\n        // Keep the highest score across occurrences\n        if (contentScore > entry.maxScore) {\n          entry.maxScore = contentScore\n          entry.candidate.contentScore = contentScore\n          entry.candidate.file = relPath\n          entry.candidate.line = extraction.line\n          entry.candidate.column = extraction.column\n          entry.candidate.context = mappedContext\n          entry.candidate.surrounding = extraction.surrounding\n        }\n      }\n    }\n  }\n\n  // Build sorted unique candidates (highest score first)\n  const allUniqueCandidates = [...uniqueMap.values()]\n    .map(e => e.candidate)\n    .toSorted((a, b) => b.contentScore - a.contentScore)\n\n  // Build duplicate groups (only count >= 2), sorted by count descending\n  const duplicates: DuplicateGroup[] = [...dupeMap.entries()]\n    .filter(([, occurrences]) => occurrences.length >= 2)\n    .map(([value, occurrences]) => ({ value, count: occurrences.length, occurrences }))\n    .toSorted((a, b) => b.count - a.count)\n\n  // Pagination on unique candidates\n  const uniqueCount = allUniqueCandidates.length\n  const paginated = allUniqueCandidates.slice(offset, offset + limit)\n  const hasMore = uniqueCount > offset + limit\n\n  return {\n    candidates: paginated,\n    duplicates,\n    stats: {\n      files_scanned: files.length,\n      raw_strings_found: rawStringsFound,\n      skipped: skippedCount,\n      low_confidence: lowConfidenceCount,\n      unique_candidates: uniqueCount,\n      candidates_returned: paginated.length,\n      has_more: hasMore,\n      skip_reasons: skipReasons,\n    },\n  }\n}\n\n// ─── Main: scanSummary ───\n\nexport async function scanSummary(\n  projectRoot: string,\n  options?: ScanOptions,\n): Promise<ScanSummaryResult> {\n  const minLength = options?.min_length ?? DEFAULT_MIN_LENGTH\n  const maxLength = options?.max_length ?? DEFAULT_MAX_LENGTH\n  const minScore = options?.min_score ?? DEFAULT_MIN_SCORE\n\n  const scanDirs = options?.paths ?? await autoDetectSourceDirs(projectRoot)\n  const files = await discoverFiles(projectRoot, {\n    paths: scanDirs,\n    include: options?.include,\n    exclude: options?.exclude,\n  })\n\n  // Group files by directory\n  const dirFiles = new Map<string, string[]>()\n  const fileTypes: Record<string, number> = {}\n\n  for (const relPath of files) {\n    const parts = relPath.split('/')\n    const dir = parts.slice(0, -1).join('/') || '.'\n    const ext = extname(relPath)\n\n    if (!dirFiles.has(dir)) dirFiles.set(dir, [])\n    dirFiles.get(dir)!.push(relPath)\n\n    fileTypes[ext] = (fileTypes[ext] ?? 0) + 1\n  }\n\n  // Sample files per directory and count candidates\n  const byDirectory: Record<string, { files: number; candidates: number }> = {}\n  const freqMap = new Map<string, number>()\n  let totalCandidatesEstimate = 0\n\n  for (const [dir, dirFileList] of dirFiles) {\n    const totalInDir = dirFileList.length\n    const sampleFiles = dirFileList.slice(0, SUMMARY_SAMPLE_SIZE)\n    let sampleCandidates = 0\n\n    const samplePromises = sampleFiles.map(async (relPath) => {\n      const filePath = join(projectRoot, relPath)\n      const content = await readText(filePath)\n      if (!content) return []\n\n      const ext = extname(filePath)\n      return extractStrings(filePath, content, ext)\n    })\n\n    const sampleResults = await Promise.all(samplePromises)\n\n    for (const extractions of sampleResults) {\n      for (const extraction of extractions) {\n        // Apply same filters as scanCandidates\n        if (extraction.value.length < minLength || extraction.value.length > maxLength) continue\n\n        const contentScore: number = (extraction as unknown as { contentScore?: number }).contentScore ?? calculateContentScore(extraction)\n        if (contentScore < minScore) continue\n\n        sampleCandidates++\n\n        // Track frequencies for top_repeated\n        const prev = freqMap.get(extraction.value) ?? 0\n        freqMap.set(extraction.value, prev + 1)\n      }\n    }\n\n    // Estimate for full directory\n    const avgPerFile = sampleFiles.length > 0 ? sampleCandidates / sampleFiles.length : 0\n    const estimatedCandidates = Math.round(avgPerFile * totalInDir)\n\n    byDirectory[dir] = {\n      files: totalInDir,\n      candidates: estimatedCandidates,\n    }\n\n    totalCandidatesEstimate += estimatedCandidates\n  }\n\n  // Top repeated strings\n  const topRepeated = [...freqMap.entries()]\n    .filter(([, count]) => count >= 2)\n    .toSorted((a, b) => b[1] - a[1])\n    .slice(0, 20)\n    .map(([value, count]) => ({ value, count }))\n\n  return {\n    total_files: files.length,\n    total_candidates_estimate: totalCandidatesEstimate,\n    by_directory: byDirectory,\n    top_repeated: topRepeated,\n    sampling_note: `Based on first ${SUMMARY_SAMPLE_SIZE} files per directory. Counts are from sampled subset, not project-wide.`,\n    file_types: fileTypes,\n  }\n}\n"],"mappings":";;;;;AAiBA,MAAM,iBAAiB;AACvB,MAAM,eAAe;AACrB,MAAM,cAAc;AACpB,MAAM,mBAAmB;AACzB,MAAM,iBAAiB;AACvB,MAAM,cAAc;AACpB,MAAM,0BAA0B;AAChC,MAAM,gBAAgB;AACtB,MAAM,iBAAiB;AACvB,MAAM,gBAAgB;AAItB,SAAS,UAAU,KAAsB;AACvC,KAAI,kCAAkC,KAAK,IAAI,CAAE,QAAO;AACxD,KAAI,4BAA4B,KAAK,IAAI,CAAE,QAAO;AAClD,KAAI,qBAAqB,KAAK,IAAI,aAAa,CAAC,IAAI,CAAC,IAAI,SAAS,IAAI,KAAK,IAAI,SAAS,IAAI,IAAI,IAAI,SAAS,IAAI,EAC/G,QAAO;AAET,QAAO;;AAKT,MAAM,sBAAsB;AAE5B,SAAS,eAAe,OAAwB;CAC9C,MAAM,WAAW,MAAM,MAAM,CAAC,MAAM,MAAM;AAC1C,KAAI,SAAS,SAAS,EAAG,QAAO;CAChC,IAAI,UAAU;AACd,MAAK,MAAM,OAAO,SAChB,KAAI,oBAAoB,KAAK,IAAI,CAAE;AAErC,QAAO,UAAU,SAAS,UAAU;;AAGtC,SAAS,mBAAmB,OAAwB;CAClD,MAAM,UAAU,MAAM,MAAM;AAC5B,KAAI,QAAQ,SAAS,IAAI,CAAE,QAAO;AAClC,QAAO,oBAAoB,KAAK,QAAQ;;AAK1C,MAAM,2BAA2B,IAAI,IAAI;CACvC;CAAK;CAAW;CAAU;CAAa;CACvC;CAAS;CACT;CAAkB;CAAmB;CACrC;CAAoB;CAAqB;CACzC;CAAa;CACd,CAAC;AAEF,MAAM,uBAAuB,IAAI,IAAI;CACnC;CAAO;CAAQ;CAAU;CAAQ;CAAQ;CAAY;CACrD;CAAW;CAAK;CAAQ;CAAO;CAAU;CAAY;CACrD;CAAW;CAAkB;CAAkB;CAC/C;CAAU;CAAW;CAAoB;CAC1C,CAAC;AAIF,MAAM,iBAAiB,IAAI,IAAI;CAC7B;CAAK;CAAM;CAAQ;CAAa;CAAiB;CAClD,CAAC;AAEF,MAAM,iBAAiB,IAAI,IAAI,CAC7B,QAAQ,QACT,CAAC;AAIF,MAAM,0BAA0B,IAAI,IAAI;CAEtC;CAAS;CAAO;CAAe;CAAS;CAAW;CACnD;CAAQ;CAAa;CAAW;CAEhC;CAAc;CAAoB;CAClC;CAAwB;CAExB;CAAsB;CAAqB;CAE3C;CAAW;CAAc;CAAW;CAAQ;CAC5C;CAAe;CAAiB;CAChC;CAAgB;CAAe;CAAc;CAC7C;CAAgB;CACjB,CAAC;AAIF,MAAM,0BAA0B,IAAI,IAAI;CACtC;CAAS;CAAS;CAAe;CAAQ;CAAW;CACpD;CAAW;CAAW;CAAW;CAAc;CAAY;CAC3D;CAAQ;CAAY;CAAgB;CAAkB;CACvD,CAAC;;;;;;;;AAWF,SAAgB,WAAW,KAAqC;AAG9D,KAAI,IAAI,YAAY,cAAe,QAAO;AAC1C,KAAI,IAAI,YAAY,kBAAmB,QAAO;AAC9C,KAAI,IAAI,YAAY,YAAa,QAAO;AACxC,KAAI,IAAI,YAAY,mBAAoB,QAAO;AAC/C,KAAI,IAAI,YAAY,eAAgB,QAAO;AAC3C,KAAI,IAAI,YAAY,iBAAkB,QAAO;AAC7C,KAAI,IAAI,YAAY,cAAe,QAAO;CAE1C,MAAM,IAAI,IAAI;AAId,KAAI,EAAE,UAAU,EAAG,QAAO;AAC1B,KAAI,QAAQ,KAAK,EAAE,CAAE,QAAO;AAC5B,KAAI,eAAe,KAAK,EAAE,CAAE,QAAO;AACnC,KAAI,aAAa,KAAK,EAAE,CAAE,QAAO;AACjC,KAAI,YAAY,KAAK,EAAE,CAAE,QAAO;AAChC,KAAI,EAAE,WAAW,KAAK,CAAE,QAAO;AAI/B,KAAI,YAAY,KAAK,EAAE,CAAE,QAAO;AAIhC,KAAI,UAAU,EAAE,CAAE,QAAO;AAIzB,KAAI,eAAe,EAAE,CAAE,QAAO;AAC9B,KAAI,mBAAmB,EAAE,CAAE,QAAO;AAIlC,KAAI,EAAE,SAAS,KAAK,iBAAiB,KAAK,EAAE,CAAE,QAAO;AACrD,KAAI,eAAe,KAAK,EAAE,CAAE,QAAO;AACnC,KAAI,IAAI,mBAAmB,KAAA,KAAa,yBAAyB,IAAI,IAAI,eAAe,CAAE,QAAO;AACjG,KAAI,IAAI,YAAY,wBAAwB,qBAAqB,IAAI,IAAI,OAAO,CAAE,QAAO;AAIzF,KAAI,EAAE,WAAW,UAAU,CAAE,QAAO;AAIpC,KAAI,eAAe,KAAK,EAAE,CAAE,QAAO;AAInC,KAAI,IAAI,YAAY,uBAAuB,eAAe,IAAI,IAAI,OAAO;MAEnE,sBAAsB,KAAK,EAAE,CAAE,QAAO;;AAG5C,KAAI,IAAI,YAAY,uBAAuB,eAAe,IAAI,IAAI,OAAO,CACvE,QAAO;AAOT,KAAI,IAAI,YAAY,mBAAmB,IAAI,YAAY;MACjD,wBAAwB,KAAK,EAAE,IAAI,EAAE,SAAS,GAChD,QAAO;;AAMX,KAAI,cAAc,KAAK,EAAE,IAAI,EAAE,SAAS,IAAI,IAAI,EAAE,SAAS,EACzD,QAAO;AAGT,QAAO;;;;;;;;;AAYT,SAAgB,sBAAsB,KAA8B;CAClE,IAAI,QAAQ;AAKZ,KAAI,IAAI,YAAY,mBAAmB,IAAI,YAAY,WACrD,UAAS;AAIX,KAAI,IAAI,YAAY,wBAAwB,IAAI,YAAY,gBAC1D,KAAI,IAAI,kBAAkB,wBAAwB,IAAI,IAAI,eAAe,CACvE,UAAS;KAET,UAAS;AAKb,KAAI,IAAI,YAAY;MACd,IAAI,kBAAkB,wBAAwB,IAAI,IAAI,eAAe,CACvE,UAAS;;CAOb,MAAM,YAAY,IAAI,MAAM,MAAM,MAAM,CAAC;AACzC,KAAI,aAAa,EAAG,UAAS;UACpB,cAAc,EAAG,UAAS;AAGnC,KAAI,WAAW,KAAK,IAAI,MAAM,CAAE,UAAS;AAGzC,KAAI,kBAAkB,KAAK,IAAI,MAAM,CAAE,UAAS;AAGhD,KAAI,SAAS,KAAK,IAAI,MAAM,IAAI,QAAQ,KAAK,IAAI,MAAM,CAAE,UAAS;AAGlE,KAAI,cAAc,KAAK,IAAI,MAAM,CAAE,UAAS;AAG5C,KAAI,IAAI,MAAM,SAAS,IAAI,IAAI,CAAC,IAAI,MAAM,SAAS,IAAI,CAAE,UAAS;AAElE,QAAO,KAAK,IAAI,GAAG,KAAK,IAAI,GAAG,MAAM,CAAC;;;;;;;;;AAYxC,SAAgB,eACd,SACA,WAAmB,IACF;CACjB,MAAM,aAAgC,EAAE;CACxC,MAAM,cAAsC,EAAE;CAC9C,IAAI,UAAU;CACd,IAAI,gBAAgB;AAEpB,MAAK,MAAM,OAAO,SAAS;EAEzB,MAAM,aAAa,WAAW,IAAI;AAClC,MAAI,YAAY;AACd;AACA,eAAY,eAAe,YAAY,eAAe,KAAK;AAC3D;;EAIF,MAAM,eAAe,sBAAsB,IAAI;AAC/C,MAAI,eAAe,UAAU;AAC3B;AACA,eAAY,qBAAqB,YAAY,qBAAqB,KAAK;AACvE;;AAIA,MAAmD,eAAe;AACpE,aAAW,KAAK,IAAI;;AAGtB,QAAO;EAAE;EAAY;EAAS;EAAe;EAAa;;;;ACvS5D,MAAM,iBAAiB,IAAI,IAAI;CAAC;CAAQ;CAAQ;CAAO;CAAO;CAAO,CAAC;AACtE,MAAM,iBAAiB,IAAI,IAAI,CAAC,OAAO,CAAC;AACxC,MAAM,oBAAoB,IAAI,IAAI,CAAC,UAAU,CAAC;AAC9C,MAAM,mBAAmB,IAAI,IAAI,CAAC,SAAS,CAAC;;;;;AAQ5C,eAAe,gBAA8H;AAC3I,KAAI;AAEF,UADY,MAAM,OAAO,8BACd;SACL;AACN;;;;;;;AAQJ,eAAe,mBAA6G;AAC1H,KAAI;AAEF,UADY,MAAM,OAAO,iCACd;SACL;AACN;;;;;;;AAQJ,eAAe,kBAA4G;AACzH,KAAI;AAEF,UADY,MAAM,OAAO,gCACd;SACL;AACN;;;AAMJ,MAAM,kBAAkB;;;;;;AAOxB,SAAS,iBAAiB,SAAiB,WAAsC;CAC/E,MAAM,QAAQ,QAAQ,MAAM,KAAK;CACjC,MAAM,UAA6B,EAAE;CAGrC,MAAM,WAAW;AAEjB,MAAK,IAAI,IAAI,GAAG,IAAI,MAAM,QAAQ,KAAK;EACrC,MAAM,OAAO,MAAM;EACnB,MAAM,UAAU,KAAK,MAAM;AAG3B,MAAI,QAAQ,WAAW,KAAK,IAAI,QAAQ,WAAW,KAAK,IAAI,QAAQ,WAAW,IAAI,CAAE;AACrF,MAAI,wBAAwB,KAAK,KAAK,CAAE;EAExC,IAAI;AACJ,WAAS,YAAY;AAErB,UAAQ,QAAQ,SAAS,KAAK,KAAK,MAAM,MAAM;GAG7C,MAAM,QADM,MAAM,GACA,MAAM,GAAG,GAAG;AAC9B,OAAI,MAAM,WAAW,EAAG;GAExB,MAAM,cAAc,iBAAiB,OAAO,EAAE;AAE9C,WAAQ,KAAK;IACX;IACA,MAAM,IAAI;IACV,QAAQ,MAAM,QAAQ;IACtB,SAAS;IACT,OAAO;IACP,QAAQ;IACR;IACD,CAAC;;EAIJ,MAAM,YAAY;EAClB,IAAI;AACJ,YAAU,YAAY;AAEtB,UAAQ,WAAW,UAAU,KAAK,KAAK,MAAM,MAAM;GACjD,MAAM,OAAO,SAAS,GAAI,MAAM;AAChC,OAAI,KAAK,WAAW,EAAG;AACvB,OAAI,YAAY,KAAK,KAAK,IAAI,CAAC,WAAW,KAAK,KAAK,CAAE;AAEtD,WAAQ,KAAK;IACX,OAAO;IACP,MAAM,IAAI;IACV,QAAQ,SAAS,QAAQ;IACzB,SAAS;IACT,OAAO;IACP,QAAQ;IACR,aAAa,iBAAiB,OAAO,EAAE;IACxC,CAAC;;;AAIN,QAAO;;AAGT,SAAS,iBAAiB,OAAiB,SAAyB;CAClE,MAAM,QAAQ,KAAK,IAAI,GAAG,UAAU,EAAE;CACtC,MAAM,MAAM,KAAK,IAAI,MAAM,SAAS,GAAG,UAAU,EAAE;CAEnD,MAAM,QAAkB,EAAE;AAC1B,MAAK,IAAI,IAAI,OAAO,KAAK,KAAK,KAAK;EACjC,MAAM,OAAO,MAAM;AACnB,MAAI,SAAS,KAAA,EACX,OAAM,KAAK,KAAK;;CAIpB,MAAM,SAAS,MAAM,KAAK,KAAK;AAC/B,KAAI,OAAO,SAAS,gBAClB,QAAO,OAAO,MAAM,GAAG,gBAAgB;AAEzC,QAAO;;;;;;;;;;;;;;;;;;AAqBT,eAAsB,eACpB,UACA,SACA,KAC4B;CAC5B,MAAM,gBAAgB,IAAI,WAAW,IAAI,GAAG,MAAM,IAAI;CACtD,IAAI;AAEJ,KAAI,eAAe,IAAI,cAAc,CACnC,cAAa,SAAS,SAAS,SAAS;UAC/B,eAAe,IAAI,cAAc,EAAE;EAC5C,MAAM,WAAW,MAAM,eAAe;AACtC,MAAI,SACF,cAAa,MAAM,SAAS,SAAS,SAAS;MAE9C,cAAa,iBAAiB,SAAS,SAAS;YAEzC,kBAAkB,IAAI,cAAc,EAAE;EAC/C,MAAM,cAAc,MAAM,kBAAkB;AAC5C,MAAI,YACF,KAAI;AACF,gBAAa,MAAM,YAAY,SAAS,SAAS;UAC3C;AAEN,gBAAa,iBAAiB,SAAS,SAAS;;MAGlD,cAAa,iBAAiB,SAAS,SAAS;YAEzC,iBAAiB,IAAI,cAAc,EAAE;EAC9C,MAAM,aAAa,MAAM,iBAAiB;AAC1C,MAAI,WACF,KAAI;AACF,gBAAa,MAAM,WAAW,SAAS,SAAS;UAC1C;AAEN,gBAAa,iBAAiB,SAAS,SAAS;;MAGlD,cAAa,iBAAiB,SAAS,SAAS;OAGlD,QAAO,EAAE;CAIX,MAAM,EAAE,eAAe,eAAe,WAAW;AACjD,QAAO;;;;AC9KT,MAAM,gBAAgB;AACtB,MAAM,iBAAiB;AACvB,MAAM,qBAAqB;AAC3B,MAAM,qBAAqB;AAC3B,MAAM,oBAAoB;AAC1B,MAAM,sBAAsB;AAI5B,MAAM,cAAwD;CAC5D,iBAAiB;CACjB,sBAAsB;CACtB,YAAY;CACZ,iBAAiB;CACjB,uBAAuB;CACvB,mBAAmB;CACnB,qBAAqB;CACrB,iBAAiB;CACjB,cAAc;CACd,oBAAoB;CACpB,eAAe;CACf,SAAS;CAET,eAAe;CACf,mBAAmB;CACnB,aAAa;CACb,oBAAoB;CACpB,gBAAgB;CAChB,kBAAkB;CACnB;AAID,eAAsB,eACpB,aACA,SAC+B;CAC/B,MAAM,QAAQ,SAAS,SAAS;CAChC,MAAM,SAAS,SAAS,UAAU;CAClC,MAAM,YAAY,SAAS,cAAc;CACzC,MAAM,YAAY,SAAS,cAAc;CACzC,MAAM,WAAW,SAAS,aAAa;CAGvC,MAAM,QAAQ,MAAM,cAAc,aAAa;EAC7C,OAFe,SAAS,SAAS,MAAM,qBAAqB,YAAY;EAGxE,SAAS,SAAS;EAClB,SAAS,SAAS;EACnB,CAAC;CAIF,MAAM,eAAe,MAAM,IAAI,OAAO,YAAY;EAChD,MAAM,WAAW,KAAK,aAAa,QAAQ;EAC3C,MAAM,UAAU,MAAM,SAAS,SAAS;AACxC,MAAI,CAAC,QAAS,QAAO;GAAE;GAAS,aAAa,EAAE;GAAE;AAIjD,SAAO;GAAE;GAAS,aADE,MAAM,eAAe,UAAU,SADvC,QAAQ,SAAS,CACmC;GACjC;GAC/B;CAEF,MAAM,cAAc,MAAM,QAAQ,IAAI,aAAa;CAInD,IAAI,kBAAkB;CACtB,IAAI,eAAe;CACnB,IAAI,qBAAqB;CACzB,MAAM,cAAsC,EAAE;CAG9C,MAAM,4BAAY,IAAI,KAGlB;CAGJ,MAAM,0BAAU,IAAI,KAAoD;AAExE,MAAK,MAAM,EAAE,SAAS,iBAAiB,aAAa;AAClD,qBAAmB,YAAY;AAE/B,OAAK,MAAM,cAAc,aAAa;AAEpC,OAAI,WAAW,MAAM,SAAS,aAAa,WAAW,MAAM,SAAS,WAAW;AAC9E;AACA,gBAAY,oBAAoB,YAAY,oBAAoB,KAAK;AACrE;;GAQF,MAAM,eAAwB,WAAoD,gBAAgB,sBAAsB,WAAW;AAEnI,OAAI,eAAe,UAAU;AAC3B;AACA,gBAAY,qBAAqB,YAAY,qBAAqB,KAAK;AACvE;;GAGF,MAAM,gBAAgB,YAAY,WAAW;GAC7C,MAAM,MAAM;IAAE,MAAM;IAAS,MAAM,WAAW;IAAM;AAGpD,OAAI,CAAC,QAAQ,IAAI,WAAW,MAAM,CAChC,SAAQ,IAAI,WAAW,OAAO,EAAE,CAAC;AAEnC,WAAQ,IAAI,WAAW,MAAM,CAAE,KAAK,IAAI;AAGxC,OAAI,CAAC,UAAU,IAAI,WAAW,MAAM,CAClC,WAAU,IAAI,WAAW,OAAO;IAC9B,WAAW;KACT,MAAM;KACN,MAAM,WAAW;KACjB,QAAQ,WAAW;KACnB,OAAO,WAAW;KAClB,SAAS;KACT,aAAa,WAAW;KACxB;KACA,aAAa,CAAC,IAAI;KACnB;IACD,UAAU;IACX,CAAC;QACG;IACL,MAAM,QAAQ,UAAU,IAAI,WAAW,MAAM;AAC7C,UAAM,UAAU,YAAY,KAAK,IAAI;AAErC,QAAI,eAAe,MAAM,UAAU;AACjC,WAAM,WAAW;AACjB,WAAM,UAAU,eAAe;AAC/B,WAAM,UAAU,OAAO;AACvB,WAAM,UAAU,OAAO,WAAW;AAClC,WAAM,UAAU,SAAS,WAAW;AACpC,WAAM,UAAU,UAAU;AAC1B,WAAM,UAAU,cAAc,WAAW;;;;;CAOjD,MAAM,sBAAsB,CAAC,GAAG,UAAU,QAAQ,CAAC,CAChD,KAAI,MAAK,EAAE,UAAU,CACrB,UAAU,GAAG,MAAM,EAAE,eAAe,EAAE,aAAa;CAGtD,MAAM,aAA+B,CAAC,GAAG,QAAQ,SAAS,CAAC,CACxD,QAAQ,GAAG,iBAAiB,YAAY,UAAU,EAAE,CACpD,KAAK,CAAC,OAAO,kBAAkB;EAAE;EAAO,OAAO,YAAY;EAAQ;EAAa,EAAE,CAClF,UAAU,GAAG,MAAM,EAAE,QAAQ,EAAE,MAAM;CAGxC,MAAM,cAAc,oBAAoB;CACxC,MAAM,YAAY,oBAAoB,MAAM,QAAQ,SAAS,MAAM;CACnE,MAAM,UAAU,cAAc,SAAS;AAEvC,QAAO;EACL,YAAY;EACZ;EACA,OAAO;GACL,eAAe,MAAM;GACrB,mBAAmB;GACnB,SAAS;GACT,gBAAgB;GAChB,mBAAmB;GACnB,qBAAqB,UAAU;GAC/B,UAAU;GACV,cAAc;GACf;EACF;;AAKH,eAAsB,YACpB,aACA,SAC4B;CAC5B,MAAM,YAAY,SAAS,cAAc;CACzC,MAAM,YAAY,SAAS,cAAc;CACzC,MAAM,WAAW,SAAS,aAAa;CAGvC,MAAM,QAAQ,MAAM,cAAc,aAAa;EAC7C,OAFe,SAAS,SAAS,MAAM,qBAAqB,YAAY;EAGxE,SAAS,SAAS;EAClB,SAAS,SAAS;EACnB,CAAC;CAGF,MAAM,2BAAW,IAAI,KAAuB;CAC5C,MAAM,YAAoC,EAAE;AAE5C,MAAK,MAAM,WAAW,OAAO;EAE3B,MAAM,MADQ,QAAQ,MAAM,IAAI,CACd,MAAM,GAAG,GAAG,CAAC,KAAK,IAAI,IAAI;EAC5C,MAAM,MAAM,QAAQ,QAAQ;AAE5B,MAAI,CAAC,SAAS,IAAI,IAAI,CAAE,UAAS,IAAI,KAAK,EAAE,CAAC;AAC7C,WAAS,IAAI,IAAI,CAAE,KAAK,QAAQ;AAEhC,YAAU,QAAQ,UAAU,QAAQ,KAAK;;CAI3C,MAAM,cAAqE,EAAE;CAC7E,MAAM,0BAAU,IAAI,KAAqB;CACzC,IAAI,0BAA0B;AAE9B,MAAK,MAAM,CAAC,KAAK,gBAAgB,UAAU;EACzC,MAAM,aAAa,YAAY;EAC/B,MAAM,cAAc,YAAY,MAAM,GAAG,oBAAoB;EAC7D,IAAI,mBAAmB;EAEvB,MAAM,iBAAiB,YAAY,IAAI,OAAO,YAAY;GACxD,MAAM,WAAW,KAAK,aAAa,QAAQ;GAC3C,MAAM,UAAU,MAAM,SAAS,SAAS;AACxC,OAAI,CAAC,QAAS,QAAO,EAAE;AAGvB,UAAO,eAAe,UAAU,SADpB,QAAQ,SAAS,CACgB;IAC7C;EAEF,MAAM,gBAAgB,MAAM,QAAQ,IAAI,eAAe;AAEvD,OAAK,MAAM,eAAe,cACxB,MAAK,MAAM,cAAc,aAAa;AAEpC,OAAI,WAAW,MAAM,SAAS,aAAa,WAAW,MAAM,SAAS,UAAW;AAGhF,QAD8B,WAAoD,gBAAgB,sBAAsB,WAAW,IAChH,SAAU;AAE7B;GAGA,MAAM,OAAO,QAAQ,IAAI,WAAW,MAAM,IAAI;AAC9C,WAAQ,IAAI,WAAW,OAAO,OAAO,EAAE;;EAK3C,MAAM,aAAa,YAAY,SAAS,IAAI,mBAAmB,YAAY,SAAS;EACpF,MAAM,sBAAsB,KAAK,MAAM,aAAa,WAAW;AAE/D,cAAY,OAAO;GACjB,OAAO;GACP,YAAY;GACb;AAED,6BAA2B;;CAI7B,MAAM,cAAc,CAAC,GAAG,QAAQ,SAAS,CAAC,CACvC,QAAQ,GAAG,WAAW,SAAS,EAAE,CACjC,UAAU,GAAG,MAAM,EAAE,KAAK,EAAE,GAAG,CAC/B,MAAM,GAAG,GAAG,CACZ,KAAK,CAAC,OAAO,YAAY;EAAE;EAAO;EAAO,EAAE;AAE9C,QAAO;EACL,aAAa,MAAM;EACnB,2BAA2B;EAC3B,cAAc;EACd,cAAc;EACd,eAAe,kBAAkB,oBAAoB;EACrD,YAAY;EACb"}

package/dist/{server-CjK63H6X.mjs → server-DfWRpzdP.mjs}

@@ -6,20 +6,124 @@ import { t as readContext } from "./context-DIDt_TxX.mjs";
 import { t as detectStack } from "./detect-DinOQAmK.mjs";
 import { n as cleanupMergedBranches, t as checkBranchHealth } from "./branch-lifecycle-wGJg25LT.mjs";
 import { n as listTemplates, t as getTemplate } from "./templates-bRPzYq4M.mjs";
-import { n as createTransaction, r as ensureContentBranch, t as buildBranchName } from "./transaction-B0V0_SE2.mjs";
+import { i as mergeBranch, n as createTransaction, r as ensureContentBranch, t as buildBranchName } from "./transaction-C_Pvux78.mjs";
 import { t as validateProject } from "./validator-SQYUV2JP.mjs";
 import { t as buildGraph } from "./graph-builder-tM7zj-SX.mjs";
-import { n as scanCandidates, r as scanSummary } from "./scanner-DPrbU-oK.mjs";
-import { n as applyExtract, r as applyReuse } from "./apply-manager-Cf5OWOJ7.mjs";
+import { n as scanSummary, t as scanCandidates } from "./scanner-CQOxARUU.mjs";
+import { n as applyExtract, r as applyReuse } from "./apply-manager-BkPMfpsN.mjs";
 import { join } from "node:path";
 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
 import { z } from "zod";
 import { appendFile, readFile, writeFile } from "node:fs/promises";
 import { CONTENTRAIN_BRANCH } from "@contentrain/types";
 import { simpleGit } from "simple-git";
+//#region src/tools/annotations.ts
+/**
+* Centralized tool annotations registry.
+* MCP clients use these hints to distinguish read-only vs. write vs. destructive tools.
+*/
+const TOOL_ANNOTATIONS = {
+	contentrain_status: {
+		title: "Project Status",
+		readOnlyHint: true,
+		destructiveHint: false,
+		idempotentHint: true
+	},
+	contentrain_describe: {
+		title: "Describe Model",
+		readOnlyHint: true,
+		destructiveHint: false,
+		idempotentHint: true
+	},
+	contentrain_describe_format: {
+		title: "Describe Format",
+		readOnlyHint: true,
+		destructiveHint: false,
+		idempotentHint: true
+	},
+	contentrain_init: {
+		title: "Initialize Project",
+		readOnlyHint: false,
+		destructiveHint: false,
+		idempotentHint: false
+	},
+	contentrain_scaffold: {
+		title: "Scaffold Template",
+		readOnlyHint: false,
+		destructiveHint: false,
+		idempotentHint: false
+	},
+	contentrain_model_save: {
+		title: "Save Model",
+		readOnlyHint: false,
+		destructiveHint: false,
+		idempotentHint: true
+	},
+	contentrain_model_delete: {
+		title: "Delete Model",
+		readOnlyHint: false,
+		destructiveHint: true,
+		idempotentHint: false
+	},
+	contentrain_content_save: {
+		title: "Save Content",
+		readOnlyHint: false,
+		destructiveHint: false,
+		idempotentHint: true
+	},
+	contentrain_content_delete: {
+		title: "Delete Content",
+		readOnlyHint: false,
+		destructiveHint: true,
+		idempotentHint: false
+	},
+	contentrain_content_list: {
+		title: "List Content",
+		readOnlyHint: true,
+		destructiveHint: false,
+		idempotentHint: true
+	},
+	contentrain_validate: {
+		title: "Validate Project",
+		readOnlyHint: false,
+		destructiveHint: false,
+		idempotentHint: true
+	},
+	contentrain_submit: {
+		title: "Submit Branches",
+		readOnlyHint: false,
+		destructiveHint: false,
+		idempotentHint: true
+	},
+	contentrain_merge: {
+		title: "Merge Branch",
+		readOnlyHint: false,
+		destructiveHint: false,
+		idempotentHint: false
+	},
+	contentrain_scan: {
+		title: "Scan Source Code",
+		readOnlyHint: true,
+		destructiveHint: false,
+		idempotentHint: true
+	},
+	contentrain_apply: {
+		title: "Apply Normalize",
+		readOnlyHint: false,
+		destructiveHint: false,
+		idempotentHint: false
+	},
+	contentrain_bulk: {
+		title: "Bulk Operations",
+		readOnlyHint: false,
+		destructiveHint: false,
+		idempotentHint: false
+	}
+};
+//#endregion
 //#region src/tools/context.ts
 function registerContextTools(server, projectRoot) {
-	server.tool("contentrain_status", "Get full project status (read-only). Returns config, models, context. Do NOT manually edit .contentrain/ based on this output.", {}, async () => {
+	server.tool("contentrain_status", "Get full project status (read-only). Returns config, models, context. Do NOT manually edit .contentrain/ based on this output.", {}, TOOL_ANNOTATIONS["contentrain_status"], async () => {
 		if (!await pathExists(join(contentrainDir(projectRoot), "config.json"))) {
 			const detectedStack = await detectStack(projectRoot);
 			return { content: [{
@@ -85,7 +189,7 @@ function registerContextTools(server, projectRoot) {
 		model: z.string().describe("Model ID (e.g. \"blog-post\", \"hero\")"),
 		include_sample: z.boolean().optional().default(false).describe("Include one sample entry"),
 		locale: z.string().optional().describe("Locale for sample content (default: config default locale)")
-	}, async ({ model: modelId, include_sample, locale }) => {
+	}, TOOL_ANNOTATIONS["contentrain_describe"], async ({ model: modelId, include_sample, locale }) => {
 		const modelDef = await readModel(projectRoot, modelId);
 		if (!modelDef) return {
 			content: [{
@@ -120,7 +224,7 @@ function registerContextTools(server, projectRoot) {
 			text: JSON.stringify(result, null, 2)
 		}] };
 	});
-	server.tool("contentrain_describe_format", "Describes the Contentrain content file format for any language/platform. Returns a comprehensive specification of the file structure, JSON formats, markdown conventions, meta files, and locale strategies.", {}, async () => {
+	server.tool("contentrain_describe_format", "Describes the Contentrain content file format for any language/platform. Returns a comprehensive specification of the file structure, JSON formats, markdown conventions, meta files, and locale strategies.", {}, TOOL_ANNOTATIONS["contentrain_describe_format"], async () => {
 		return { content: [{
 			type: "text",
 			text: JSON.stringify({
@@ -331,7 +435,7 @@ function registerSetupTools(server, projectRoot) {
 		stack: z.string().optional().describe("Framework stack (nuxt, next, astro, svelte, react-vite, other). Auto-detected if omitted."),
 		locales: z.array(z.string()).optional().describe("Supported locales. Default: [\"en\"]"),
 		domains: z.array(z.string()).optional().describe("Content domains. Default: auto-suggested")
-	}, async ({ stack, locales, domains }) => {
+	}, TOOL_ANNOTATIONS["contentrain_init"], async ({ stack, locales, domains }) => {
 		if (await pathExists(join(contentrainDir(projectRoot), "config.json"))) return {
 			content: [{
 				type: "text",
@@ -457,7 +561,7 @@ function registerSetupTools(server, projectRoot) {
 		template: z.string().describe("Template ID: blog, landing, docs, ecommerce, saas, i18n, mobile"),
 		locales: z.array(z.string()).optional().describe("Override locales"),
 		with_sample_content: z.boolean().optional().default(true).describe("Include sample content (default: true)")
-	}, async ({ template: templateId, locales, with_sample_content }) => {
+	}, TOOL_ANNOTATIONS["contentrain_scaffold"], async ({ template: templateId, locales, with_sample_content }) => {
 		const config = await readConfig(projectRoot);
 		if (!config) return {
 			content: [{
@@ -636,7 +740,7 @@ function registerModelTools(server, projectRoot) {
 			"directory",
 			"none"
 		]).optional().describe("How locale is encoded in file names. Default: \"file\"")
-	}, async (input) => {
+	}, TOOL_ANNOTATIONS["contentrain_model_save"], async (input) => {
 		const config = await readConfig(projectRoot);
 		if (!config) return {
 			content: [{
@@ -747,7 +851,7 @@ function registerModelTools(server, projectRoot) {
 	server.tool("contentrain_model_delete", "Delete a model and its content/meta. Changes are auto-committed to git — do NOT manually edit .contentrain/ files.", {
 		model: z.string().describe("Model ID to delete"),
 		confirm: z.literal(true).describe("Must be true to confirm deletion")
-	}, async ({ model: modelId }) => {
+	}, TOOL_ANNOTATIONS["contentrain_model_delete"], async ({ model: modelId }) => {
 		if (!await readConfig(projectRoot)) return {
 			content: [{
 				type: "text",
@@ -840,7 +944,7 @@ function registerContentTools(server, projectRoot) {
 			publish_at: z.string().optional().describe("ISO 8601 date for scheduled publishing (stored in meta)"),
 			expire_at: z.string().optional().describe("ISO 8601 date for scheduled expiry (stored in meta, must be after publish_at)")
 		})).describe("Content entries to save")
-	}, async (input) => {
+	}, TOOL_ANNOTATIONS["contentrain_content_save"], async (input) => {
 		const config = await readConfig(projectRoot);
 		if (!config) return {
 			content: [{
@@ -966,7 +1070,7 @@ function registerContentTools(server, projectRoot) {
 		locale: z.string().optional().describe("Locale code"),
 		keys: z.array(z.string()).optional().describe("Dictionary only: specific keys to remove. Omit to delete entire locale file."),
 		confirm: z.literal(true).describe("Must be true to confirm deletion")
-	}, async (input) => {
+	}, TOOL_ANNOTATIONS["contentrain_content_delete"], async (input) => {
 		if (!await readConfig(projectRoot)) return {
 			content: [{
 				type: "text",
@@ -1048,7 +1152,7 @@ function registerContentTools(server, projectRoot) {
 		resolve: z.boolean().optional().describe("Resolve relation fields to actual data"),
 		limit: z.number().optional().describe("Max entries to return"),
 		offset: z.number().optional().describe("Skip N entries")
-	}, async (input) => {
+	}, TOOL_ANNOTATIONS["contentrain_content_list"], async (input) => {
 		const config = await readConfig(projectRoot);
 		if (!config) return {
 			content: [{
@@ -1094,7 +1198,7 @@ function registerWorkflowTools(server, projectRoot) {
 	server.tool("contentrain_validate", "Validate project content against model schemas. Detects required field violations, type mismatches, broken relations, secret leaks, i18n parity issues, and more. If fix:true, auto-fixes structural issues (canonical sort, orphan meta, missing locale files) — do NOT manually edit .contentrain/ files.", {
 		model: z.string().optional().describe("Model ID to validate (omit for all models)"),
 		fix: z.boolean().optional().describe("Auto-fix structural issues (canonical sort, orphan meta, missing locale files). Default: false")
-	}, async (input) => {
+	}, TOOL_ANNOTATIONS["contentrain_validate"], async (input) => {
 		if (!await readConfig(projectRoot)) return {
 			content: [{
 				type: "text",
@@ -1190,7 +1294,7 @@ function registerWorkflowTools(server, projectRoot) {
 	server.tool("contentrain_submit", "Push contentrain/* branches to remote. MCP is push-only — PR creation is handled by the platform. Do NOT manually push or create PRs.", {
 		branches: z.array(z.string()).optional().describe("Specific branch names to push (omit for all contentrain/* branches)"),
 		message: z.string().optional().describe("Optional message for the push operation")
-	}, async (input) => {
+	}, TOOL_ANNOTATIONS["contentrain_submit"], async (input) => {
 		const config = await readConfig(projectRoot);
 		if (!config) return {
 			content: [{
@@ -1271,6 +1375,7 @@ function registerWorkflowTools(server, projectRoot) {
 			const cleanup = await cleanupMergedBranches(projectRoot);
 			const nextSteps = [];
 			if (pushed.length > 0) nextSteps.push("Create PRs on your git platform for review");
+			if (pushed.length > 0) nextSteps.push("For merge: use contentrain_merge or direct user to http://localhost:3333/branches");
 			if (errors.length > 0) nextSteps.push("Fix push errors and retry");
 			if (cleanup.remaining >= 50) nextSteps.push(`Warning: ${cleanup.remaining} active contentrain branches. Consider reviewing old branches.`);
 			return { content: [{
@@ -1298,6 +1403,53 @@ function registerWorkflowTools(server, projectRoot) {
 			};
 		}
 	});
+	server.tool("contentrain_merge", "Merge a review-mode branch into contentrain. Local git operation — no external platform needed. Merges the feature branch into the contentrain branch, advances the base branch via update-ref, and selectively syncs .contentrain/ files to the working tree.", {
+		branch: z.string().describe("Branch name to merge (e.g. cr/normalize/extract/...)"),
+		confirm: z.literal(true).describe("Must be true to confirm the merge")
+	}, TOOL_ANNOTATIONS["contentrain_merge"], async (input) => {
+		if (!await readConfig(projectRoot)) return {
+			content: [{
+				type: "text",
+				text: JSON.stringify({ error: "Project not initialized. Run contentrain_init first." })
+			}],
+			isError: true
+		};
+		try {
+			if (!(await simpleGit(projectRoot).branchLocal()).all.includes(input.branch)) return {
+				content: [{
+					type: "text",
+					text: JSON.stringify({
+						error: `Branch "${input.branch}" not found locally.`,
+						next_steps: ["Check branch name with git branch -l", "Use contentrain_submit to push first if needed"]
+					})
+				}],
+				isError: true
+			};
+			const result = await mergeBranch(projectRoot, input.branch);
+			return { content: [{
+				type: "text",
+				text: JSON.stringify({
+					status: "merged",
+					action: result.action,
+					commit: result.commit,
+					sync: result.sync,
+					next_steps: [
+						"Run `npx contentrain generate` to update SDK client",
+						"Run contentrain_validate to verify content integrity",
+						result.sync.skipped.length > 0 ? `${result.sync.skipped.length} file(s) skipped due to local changes — resolve manually` : ""
+					].filter(Boolean)
+				}, null, 2)
+			}] };
+		} catch (error) {
+			return {
+				content: [{
+					type: "text",
+					text: JSON.stringify({ error: `Merge failed: ${error instanceof Error ? error.message : String(error)}` })
+				}],
+				isError: true
+			};
+		}
+	});
 }
 //#endregion
 //#region src/tools/normalize.ts
@@ -1315,7 +1467,7 @@ function registerNormalizeTools(server, projectRoot) {
 		offset: z.number().optional().describe("Candidates mode: pagination offset. Default: 0"),
 		min_length: z.number().optional().describe("Candidates mode: minimum string length. Default: 2"),
 		max_length: z.number().optional().describe("Candidates mode: maximum string length. Default: 500")
-	}, async (input) => {
+	}, TOOL_ANNOTATIONS["contentrain_scan"], async (input) => {
 		if (!await readConfig(projectRoot)) return {
 			content: [{
 				type: "text",
@@ -1441,7 +1593,7 @@ function registerNormalizeTools(server, projectRoot) {
 			new_expression: z.string().describe("Replacement expression (agent-determined)"),
 			import_statement: z.string().optional().describe("Import to add if needed")
 		})).optional().describe("Reuse mode: patches to apply (max 100)")
-	}, async (input) => {
+	}, TOOL_ANNOTATIONS["contentrain_apply"], async (input) => {
 		if (!await readConfig(projectRoot)) return {
 			content: [{
 				type: "text",
@@ -1562,7 +1714,7 @@ function registerBulkTools(server, projectRoot) {
 			"archived"
 		]).optional().describe("New status for update_status"),
 		confirm: z.boolean().optional().describe("Must be true for delete_entries")
-	}, async (input) => {
+	}, TOOL_ANNOTATIONS["contentrain_bulk"], async (input) => {
 		const config = await readConfig(projectRoot);
 		if (!config) return {
 			content: [{
@@ -1860,7 +2012,7 @@ function registerBulkTools(server, projectRoot) {
 }
 //#endregion
 //#region package.json
-var version = "1.1.2";
+var version = "1.2.0";
 //#endregion
 //#region src/server.ts
 function createServer(projectRoot) {
@@ -1880,4 +2032,4 @@ function createServer(projectRoot) {
 //#endregion
 export { createServer as t };
 
-//# sourceMappingURL=server-CjK63H6X.mjs.map
+//# sourceMappingURL=server-DfWRpzdP.mjs.map