@theglitchking/babel-fish 1.0.2 → 2.0.1

package/.claude/project-map/generate.py

@@ -0,0 +1,1494 @@
+#!/usr/bin/env python3
+"""
+generate.py — Babel Fish
+Stack-agnostic introspection script for the babel-fish codebase mapper plugin.
+Produces a split-section project map under .claude/project-map/sections/
+
+Usage:
+    python generate.py [--force] [--project-root PATH] [--stack-json PATH]
+"""
+from __future__ import annotations
+
+import ast
+import argparse
+import hashlib
+import json
+import os
+import re
+import subprocess
+import sys
+from datetime import datetime
+from pathlib import Path
+from typing import Any
+
+# ── Try optional deps ────────────────────────────────────────────────────────
+try:
+    import yaml
+    HAS_YAML = True
+except ImportError:
+    HAS_YAML = False
+
+# ── Paths ────────────────────────────────────────────────────────────────────
+SCRIPT_DIR   = Path(__file__).parent
+MAP_DIR      = SCRIPT_DIR
+SECTIONS_DIR = MAP_DIR / "sections"
+CHECKSUMS    = MAP_DIR / "checksums.json"
+LEARNED_VOC  = MAP_DIR / "learned-vocabulary.json"
+
+# Resolve project root: two levels up from .claude/project-map/
+PROJECT_ROOT = MAP_DIR.parent.parent
+
+SECTIONS_DIR.mkdir(parents=True, exist_ok=True)
+
+# ── Secrets guard ────────────────────────────────────────────────────────────
+SECRET_PATTERNS = re.compile(
+    r'(?i)(password|secret|token|api_key|apikey|private_key|auth_token|'
+    r'access_key|secret_key|client_secret|db_pass|database_password|'
+    r'stripe_key|twilio_auth|sendgrid_key|aws_secret)\s*[=:]\s*\S+'
+)
+
+def redact_secrets(text: str) -> str:
+    return SECRET_PATTERNS.sub(r'[REDACTED]', text)
+
+
+# ── Checksum logic ───────────────────────────────────────────────────────────
+WATCHED_EXTENSIONS = {
+    '.py', '.ts', '.tsx', '.js', '.jsx', '.go', '.java', '.kt',
+    '.yaml', '.yml', '.toml', '.json', '.prisma', '.sql', '.env',
+}
+WATCHED_NAMES = {
+    'docker-compose.yml', 'docker-compose.yaml', 'docker-compose.dev.yml',
+    'package.json', 'requirements.txt', 'pyproject.toml', 'go.mod',
+    'Cargo.toml', 'pom.xml', 'Gemfile', 'Makefile',
+}
+IGNORE_DIRS = {
+    '.git', 'node_modules', '__pycache__', '.venv', 'venv', 'env',
+    'dist', 'build', '.next', '.nuxt', 'target', 'vendor', '.cache',
+    '.claude', '.planning',
+}
+
+def collect_watched_files() -> list[Path]:
+    result = []
+    for path in sorted(PROJECT_ROOT.rglob('*')):
+        if any(p in IGNORE_DIRS for p in path.parts):
+            continue
+        if path.is_file() and (path.suffix in WATCHED_EXTENSIONS or path.name in WATCHED_NAMES):
+            result.append(path)
+    return result
+
+def compute_checksum(files: list[Path]) -> str:
+    h = hashlib.sha256()
+    for f in files:
+        h.update(str(f).encode())
+        try:
+            h.update(str(f.stat().st_mtime_ns).encode())
+        except OSError:
+            pass
+    return h.hexdigest()
+
+def load_checksums() -> dict:
+    if CHECKSUMS.exists():
+        try:
+            return json.loads(CHECKSUMS.read_text())
+        except Exception:
+            pass
+    return {}
+
+def save_checksums(data: dict) -> None:
+    CHECKSUMS.write_text(json.dumps(data, indent=2))
+
+def is_unchanged(checksum: str) -> bool:
+    stored = load_checksums()
+    return stored.get('input_hash') == checksum
+
+
+# ── Stack detection (reads stack.json if present, else fallback) ─────────────
+def load_stack(stack_json_path: Path | None = None) -> dict:
+    candidates = [
+        stack_json_path,
+        MAP_DIR / "stack.json",
+        PROJECT_ROOT / ".claude" / "stack.json",
+    ]
+    for path in candidates:
+        if path and path.exists():
+            try:
+                return json.loads(path.read_text())
+            except Exception:
+                pass
+    return {
+        "name": PROJECT_ROOT.name,
+        "slug": re.sub(r'[^a-z0-9]', '-', PROJECT_ROOT.name.lower()).strip('-'),
+        "language": "unknown",
+        "framework": "unknown",
+        "database": "unknown",
+        "orm": "unknown",
+        "auth": "unknown",
+        "package_manager": "unknown",
+        "infrastructure": "none",
+        "project_root": str(PROJECT_ROOT),
+    }
+
+
+# ╔══════════════════════════════════════════════════════════════════════════╗
+# ║  PARSERS                                                                 ║
+# ╚══════════════════════════════════════════════════════════════════════════╝
+
+# ── Python / FastAPI / Django / Flask ────────────────────────────────────────
+
+class PythonRouteParser:
+    """Extract routes from Python routers using ast."""
+
+    DECORATOR_PATTERNS = re.compile(
+        r'@(router|app|api_router|blueprint)\.(get|post|put|patch|delete|head|options|websocket)\s*\('
+    )
+
+    def parse(self, files: list[Path]) -> list[dict]:
+        routes = []
+        for f in files:
+            if f.suffix != '.py':
+                continue
+            try:
+                source = f.read_text(encoding='utf-8', errors='replace')
+                tree = ast.parse(source, filename=str(f))
+                routes.extend(self._extract_routes(tree, f, source))
+            except SyntaxError:
+                pass
+        return routes
+
+    def _extract_routes(self, tree: ast.AST, filepath: Path, source: str) -> list[dict]:
+        routes = []
+        lines = source.splitlines()
+        for node in ast.walk(tree):
+            if not isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef)):
+                continue
+            for decorator in node.decorator_list:
+                info = self._parse_decorator(decorator, lines, node.lineno)
+                if info:
+                    info['file'] = str(filepath.relative_to(PROJECT_ROOT))
+                    info['line'] = node.lineno
+                    info['function'] = node.name
+                    routes.append(info)
+        return routes
+
+    def _parse_decorator(self, decorator: ast.expr, lines: list[str], lineno: int) -> dict | None:
+        if isinstance(decorator, ast.Call):
+            func = decorator.func
+            if isinstance(func, ast.Attribute) and func.attr in (
+                'get','post','put','patch','delete','head','options','websocket'
+            ):
+                method = func.attr.upper()
+                path = ''
+                if decorator.args:
+                    arg = decorator.args[0]
+                    if isinstance(arg, ast.Constant):
+                        path = str(arg.value)
+                return {'method': method, 'path': path}
+        return None
+
+
+class PythonModelParser:
+    """Extract SQLAlchemy / Django ORM models using ast."""
+
+    def parse(self, files: list[Path]) -> list[dict]:
+        models = []
+        for f in files:
+            if f.suffix != '.py':
+                continue
+            try:
+                source = f.read_text(encoding='utf-8', errors='replace')
+                tree = ast.parse(source, filename=str(f))
+                models.extend(self._extract_models(tree, f))
+            except SyntaxError:
+                pass
+        return models
+
+    def _extract_models(self, tree: ast.AST, filepath: Path) -> list[dict]:
+        models = []
+        for node in ast.walk(tree):
+            if not isinstance(node, ast.ClassDef):
+                continue
+            bases = [self._base_name(b) for b in node.bases]
+            is_model = any(
+                b in ('Base', 'Model', 'BaseModel', 'DeclarativeBase', 'AbstractModel')
+                or 'Model' in b
+                for b in bases if b
+            )
+            if not is_model:
+                continue
+            columns = self._extract_columns(node)
+            models.append({
+                'name': node.name,
+                'file': str(filepath.relative_to(PROJECT_ROOT)),
+                'line': node.lineno,
+                'bases': bases,
+                'columns': columns,
+            })
+        return models
+
+    def _base_name(self, node: ast.expr) -> str:
+        if isinstance(node, ast.Name):
+            return node.id
+        if isinstance(node, ast.Attribute):
+            return node.attr
+        return ''
+
+    def _extract_columns(self, class_node: ast.ClassDef) -> list[dict]:
+        cols = []
+        for node in ast.walk(class_node):
+            if isinstance(node, ast.AnnAssign) and isinstance(node.target, ast.Name):
+                cols.append({
+                    'name': node.target.id,
+                    'type': ast.unparse(node.annotation) if hasattr(ast, 'unparse') else '?',
+                })
+            elif isinstance(node, ast.Assign):
+                for target in node.targets:
+                    if isinstance(target, ast.Name):
+                        if isinstance(node.value, ast.Call):
+                            func_name = ''
+                            if isinstance(node.value.func, ast.Name):
+                                func_name = node.value.func.id
+                            elif isinstance(node.value.func, ast.Attribute):
+                                func_name = node.value.func.attr
+                            if func_name in ('Column', 'Field', 'CharField', 'IntegerField',
+                                             'TextField', 'BooleanField', 'ForeignKey',
+                                             'ManyToManyField', 'DateTimeField', 'mapped_column'):
+                                cols.append({'name': target.id, 'type': func_name})
+        return cols
+
+
+class PythonSchemaParser:
+    """Extract Pydantic schemas / Django serializers using ast."""
+
+    def parse(self, files: list[Path]) -> list[dict]:
+        schemas = []
+        for f in files:
+            if f.suffix != '.py':
+                continue
+            try:
+                source = f.read_text(encoding='utf-8', errors='replace')
+                tree = ast.parse(source, filename=str(f))
+                schemas.extend(self._extract_schemas(tree, f))
+            except SyntaxError:
+                pass
+        return schemas
+
+    def _extract_schemas(self, tree: ast.AST, filepath: Path) -> list[dict]:
+        schemas = []
+        for node in ast.walk(tree):
+            if not isinstance(node, ast.ClassDef):
+                continue
+            bases = [self._base_name(b) for b in node.bases]
+            is_schema = any(
+                b in ('BaseModel', 'Schema', 'Serializer', 'ModelSerializer',
+                       'TypedDict', 'NamedTuple')
+                or 'Schema' in b or 'Serializer' in b
+                for b in bases if b
+            )
+            if not is_schema:
+                continue
+            fields = [
+                n.target.id
+                for n in ast.walk(node)
+                if isinstance(n, ast.AnnAssign) and isinstance(n.target, ast.Name)
+            ]
+            schemas.append({
+                'name': node.name,
+                'file': str(filepath.relative_to(PROJECT_ROOT)),
+                'line': node.lineno,
+                'fields': fields,
+            })
+        return schemas
+
+    def _base_name(self, node: ast.expr) -> str:
+        if isinstance(node, ast.Name):
+            return node.id
+        if isinstance(node, ast.Attribute):
+            return node.attr
+        return ''
+
+
+# ── TypeScript / JavaScript ──────────────────────────────────────────────────
+
+class TSRouteParser:
+    """Extract routes from Next.js, Express, NestJS via regex."""
+
+    NEXT_APP_ROUTE   = re.compile(r'export\s+async\s+function\s+(GET|POST|PUT|PATCH|DELETE|HEAD)\s*\(')
+    NEXT_PAGES_ROUTE = re.compile(r'export\s+default\s+(?:async\s+)?function\s+handler')
+    EXPRESS_ROUTE    = re.compile(r'(?:router|app)\.(get|post|put|patch|delete)\s*\(\s*[\'"]([^\'"]+)[\'"]')
+    NEST_DECORATOR   = re.compile(r'@(Get|Post|Put|Patch|Delete)\s*\(\s*(?:[\'"]([^\'"]*)[\'"])?\s*\)')
+
+    def parse(self, files: list[Path]) -> list[dict]:
+        routes = []
+        for f in files:
+            if f.suffix not in ('.ts', '.tsx', '.js', '.jsx'):
+                continue
+            try:
+                source = f.read_text(encoding='utf-8', errors='replace')
+            except Exception:
+                continue
+            rel = str(f.relative_to(PROJECT_ROOT))
+
+            # Next.js App Router
+            for m in self.NEXT_APP_ROUTE.finditer(source):
+                line = source[:m.start()].count('\n') + 1
+                path = self._infer_next_path(f)
+                routes.append({'method': m.group(1), 'path': path, 'file': rel, 'line': line})
+
+            # Express
+            for m in self.EXPRESS_ROUTE.finditer(source):
+                line = source[:m.start()].count('\n') + 1
+                routes.append({'method': m.group(1).upper(), 'path': m.group(2), 'file': rel, 'line': line})
+
+            # NestJS
+            for m in self.NEST_DECORATOR.finditer(source):
+                line = source[:m.start()].count('\n') + 1
+                routes.append({'method': m.group(1).upper(), 'path': m.group(2) or '/', 'file': rel, 'line': line})
+
+        return routes
+
+    def _infer_next_path(self, f: Path) -> str:
+        """Infer URL path from Next.js file location."""
+        parts = f.parts
+        try:
+            # app router: after 'app/'
+            app_idx = parts.index('app') if 'app' in parts else None
+            if app_idx:
+                seg = parts[app_idx+1:]
+                seg = [s for s in seg if not s.startswith('(') and s not in ('route.ts','route.js','page.tsx','page.ts')]
+                return '/' + '/'.join(seg) if seg else '/'
+            # pages router
+            pages_idx = parts.index('pages') if 'pages' in parts else None
+            if pages_idx:
+                seg = list(parts[pages_idx+1:])
+                seg[-1] = re.sub(r'\.(ts|tsx|js|jsx)$', '', seg[-1])
+                if seg[-1] in ('index',):
+                    seg = seg[:-1]
+                return '/' + '/'.join(seg) if seg else '/'
+        except (ValueError, IndexError):
+            pass
+        return f'/{f.stem}'
+
+
+class TSModelParser:
+    """Extract Prisma schema entities and TypeORM entities via regex."""
+
+    PRISMA_MODEL  = re.compile(r'^model\s+(\w+)\s*\{([^}]+)\}', re.MULTILINE)
+    PRISMA_FIELD  = re.compile(r'^\s+(\w+)\s+(\w+)', re.MULTILINE)
+    TYPEORM_ENTITY = re.compile(r'@Entity\s*\(')
+    CLASS_NAME    = re.compile(r'class\s+(\w+)')
+
+    def parse(self, files: list[Path]) -> list[dict]:
+        models = []
+        for f in files:
+            rel = str(f.relative_to(PROJECT_ROOT))
+            # Prisma
+            if f.suffix == '.prisma':
+                try:
+                    source = f.read_text(encoding='utf-8', errors='replace')
+                    for m in self.PRISMA_MODEL.finditer(source):
+                        fields = [
+                            {'name': fm.group(1), 'type': fm.group(2)}
+                            for fm in self.PRISMA_FIELD.finditer(m.group(2))
+                        ]
+                        models.append({'name': m.group(1), 'file': rel, 'columns': fields})
+                except Exception:
+                    pass
+            # TypeORM / NestJS entities
+            elif f.suffix in ('.ts', '.tsx'):
+                try:
+                    source = f.read_text(encoding='utf-8', errors='replace')
+                    if self.TYPEORM_ENTITY.search(source):
+                        cm = self.CLASS_NAME.search(source)
+                        if cm:
+                            models.append({'name': cm.group(1), 'file': rel, 'columns': []})
+                except Exception:
+                    pass
+        return models
+
+
+# ── Go ───────────────────────────────────────────────────────────────────────
+
+class GoRouteParser:
+    GIN    = re.compile(r'(?:router|r|v\d+|api)\.(GET|POST|PUT|PATCH|DELETE)\s*\(\s*"([^"]+)"')
+    CHI    = re.compile(r'r\.(Get|Post|Put|Patch|Delete)\s*\(\s*"([^"]+)"')
+    STD    = re.compile(r'(?:mux|http)\.HandleFunc\s*\(\s*"([^"]+)"')
+
+    def parse(self, files: list[Path]) -> list[dict]:
+        routes = []
+        for f in files:
+            if f.suffix != '.go':
+                continue
+            try:
+                source = f.read_text(encoding='utf-8', errors='replace')
+            except Exception:
+                continue
+            rel = str(f.relative_to(PROJECT_ROOT))
+            for m in self.GIN.finditer(source):
+                line = source[:m.start()].count('\n') + 1
+                routes.append({'method': m.group(1), 'path': m.group(2), 'file': rel, 'line': line})
+            for m in self.CHI.finditer(source):
+                line = source[:m.start()].count('\n') + 1
+                routes.append({'method': m.group(1).upper(), 'path': m.group(2), 'file': rel, 'line': line})
+            for m in self.STD.finditer(source):
+                line = source[:m.start()].count('\n') + 1
+                routes.append({'method': 'ANY', 'path': m.group(1), 'file': rel, 'line': line})
+        return routes
+
+
+class GoModelParser:
+    STRUCT = re.compile(r'type\s+(\w+)\s+struct\s*\{([^}]+)\}', re.DOTALL)
+    FIELD  = re.compile(r'^\s+(\w+)\s+(\S+)', re.MULTILINE)
+
+    def parse(self, files: list[Path]) -> list[dict]:
+        models = []
+        for f in files:
+            if f.suffix != '.go':
+                continue
+            try:
+                source = f.read_text(encoding='utf-8', errors='replace')
+            except Exception:
+                continue
+            rel = str(f.relative_to(PROJECT_ROOT))
+            for m in self.STRUCT.finditer(source):
+                # Only include structs that look like data models (have db/json tags)
+                body = m.group(2)
+                if 'db:' in body or 'json:' in body or 'gorm:' in body:
+                    fields = [
+                        {'name': fm.group(1), 'type': fm.group(2)}
+                        for fm in self.FIELD.finditer(body)
+                        if not fm.group(1).startswith('//')
+                    ]
+                    models.append({'name': m.group(1), 'file': rel, 'columns': fields})
+        return models
+
+
+# ── Docker Compose ───────────────────────────────────────────────────────────
+
+class DockerComposeParser:
+    def parse(self) -> list[dict]:
+        services = []
+        candidates = [
+            'docker-compose.yml', 'docker-compose.yaml',
+            'docker-compose.dev.yml', 'docker-compose.development.yml',
+            'docker-compose.prod.yml', 'docker-compose.production.yml',
+        ]
+        for name in candidates:
+            path = PROJECT_ROOT / name
+            if not path.exists():
+                continue
+            try:
+                data = self._load(path)
+                if data and isinstance(data.get('services'), dict):
+                    for svc_name, svc in data['services'].items():
+                        services.append({
+                            'name': svc_name,
+                            'image': svc.get('image', svc.get('build', '(build)')),
+                            'ports': svc.get('ports', []),
+                            'depends_on': svc.get('depends_on', []),
+                            'environment_keys': list(svc.get('environment', {}).keys())
+                                if isinstance(svc.get('environment'), dict)
+                                else [],
+                            'source_file': name,
+                        })
+            except Exception:
+                pass
+        return services
+
+    def _load(self, path: Path) -> dict | None:
+        text = path.read_text(encoding='utf-8', errors='replace')
+        if HAS_YAML:
+            return yaml.safe_load(text)
+        # Regex fallback: extract service names at minimum
+        services: dict = {'services': {}}
+        in_services = False
+        indent = 0
+        for line in text.splitlines():
+            if re.match(r'^services\s*:', line):
+                in_services = True
+                continue
+            if in_services:
+                m = re.match(r'^(\s+)(\w[\w-]*)\s*:', line)
+                if m:
+                    lvl = len(m.group(1))
+                    if indent == 0:
+                        indent = lvl
+                    if lvl == indent:
+                        services['services'][m.group(2)] = {}
+        return services
+
+
+# ── Env Parser ───────────────────────────────────────────────────────────────
+
+class EnvParser:
+    SECRET_KEYS = re.compile(
+        r'(?i)(password|secret|token|key|auth|credential|private|cert)'
+    )
+
+    def parse(self) -> list[dict]:
+        entries = []
+        candidates = ['.env', '.env.example', '.env.local', '.env.development', '.env.sample']
+        for name in candidates:
+            path = PROJECT_ROOT / name
+            if not path.exists():
+                continue
+            for line in path.read_text(encoding='utf-8', errors='replace').splitlines():
+                line = line.strip()
+                if not line or line.startswith('#'):
+                    continue
+                if '=' not in line:
+                    continue
+                key, _, raw_val = line.partition('=')
+                key = key.strip()
+                val = raw_val.strip()
+                is_secret = bool(self.SECRET_KEYS.search(key))
+                entries.append({
+                    'key': key,
+                    'value': '[SECRET]' if is_secret else val[:80],
+                    'is_secret': is_secret,
+                    'source': name,
+                })
+        return entries
+
+
+# ── Migration Parser ─────────────────────────────────────────────────────────
+
+class MigrationParser:
+    def parse(self) -> list[dict]:
+        migrations = []
+        # Alembic
+        alembic_dirs = list(PROJECT_ROOT.rglob('versions'))
+        for d in alembic_dirs:
+            if not d.is_dir():
+                continue
+            for f in sorted(d.glob('*.py')):
+                try:
+                    source = f.read_text(encoding='utf-8', errors='replace')
+                    rev = re.search(r'revision\s*=\s*[\'"]([^\'"]+)[\'"]', source)
+                    down = re.search(r'down_revision\s*=\s*[\'"]?([^\'")\n]+)[\'"]?', source)
+                    msg = re.search(r'Create Date.*\n.*"""(.+?)"""', source, re.DOTALL)
+                    tables = re.findall(r'op\.(?:create_table|drop_table|add_column)\s*\(\s*[\'"]([^\'"]+)[\'"]', source)
+                    migrations.append({
+                        'revision': rev.group(1) if rev else f.stem,
+                        'down_revision': down.group(1).strip() if down else None,
+                        'tables_affected': list(set(tables)),
+                        'file': str(f.relative_to(PROJECT_ROOT)),
+                        'type': 'alembic',
+                    })
+                except Exception:
+                    pass
+
+        # SQL files
+        for f in sorted(PROJECT_ROOT.rglob('*.sql')):
+            if any(p in IGNORE_DIRS for p in f.parts):
+                continue
+            try:
+                source = f.read_text(encoding='utf-8', errors='replace')[:2000]
+                tables = re.findall(r'(?:CREATE|ALTER|DROP)\s+TABLE\s+(?:IF\s+NOT\s+EXISTS\s+)?(\w+)', source, re.IGNORECASE)
+                if tables:
+                    migrations.append({
+                        'revision': f.stem,
+                        'tables_affected': list(set(tables)),
+                        'file': str(f.relative_to(PROJECT_ROOT)),
+                        'type': 'sql',
+                    })
+            except Exception:
+                pass
+
+        return migrations
+
+
+# ── Frontend Feature Scanner ─────────────────────────────────────────────────
+
+class FrontendScanner:
+    FEATURE_DIRS = {'features', 'pages', 'views', 'screens', 'modules', 'app'}
+    COMPONENT_EXTS = {'.tsx', '.jsx'}
+    HOOK_PATTERN = re.compile(r'^use[A-Z]')
+    # Candidate bases to search for the FEATURE_DIRS, relative to PROJECT_ROOT.
+    # Covers repo-root and src/ layouts plus monorepos where the frontend is a
+    # subpackage (e.g. frontend/src/features), not just at the repo root.
+    SEARCH_PREFIXES = ('', 'src', 'frontend', 'frontend/src',
+                       'web', 'web/src', 'client', 'client/src', 'ui', 'ui/src')
+
+    def scan(self) -> list[dict]:
+        features, seen = [], set()
+        for prefix in self.SEARCH_PREFIXES:
+            base = PROJECT_ROOT / prefix if prefix else PROJECT_ROOT
+            for feat_dir_name in self.FEATURE_DIRS:
+                feat_dir = base / feat_dir_name
+                if not feat_dir.is_dir():
+                    continue
+                for entry in sorted(feat_dir.iterdir()):
+                    if not (entry.is_dir() and not entry.name.startswith('.')):
+                        continue
+                    rp = entry.resolve()
+                    if rp in seen:
+                        continue  # already counted via another prefix
+                    stats = self._scan_feature_dir(entry)
+                    # Only emit dirs that actually contain frontend components, so a
+                    # backend package named e.g. 'app/' is not misread as a feature.
+                    if stats['component_count'] + stats['hook_count'] == 0:
+                        continue
+                    seen.add(rp)
+                    stats['name'] = entry.name
+                    stats['path'] = str(entry.relative_to(PROJECT_ROOT))
+                    features.append(stats)
+        return features
+
+    def _scan_feature_dir(self, d: Path) -> dict:
+        components, hooks, stores, api_files = [], [], [], []
+        for f in d.rglob('*'):
+            if f.suffix in self.COMPONENT_EXTS:
+                if self.HOOK_PATTERN.match(f.stem):
+                    hooks.append(f.name)
+                else:
+                    components.append(f.name)
+            elif 'store' in f.name.lower() or 'slice' in f.name.lower():
+                stores.append(f.name)
+            elif 'api' in f.name.lower() or 'service' in f.name.lower() or 'client' in f.name.lower():
+                api_files.append(f.name)
+        return {
+            'components': components,
+            'hooks': hooks,
+            'stores': stores,
+            'api_files': api_files,
+            'component_count': len(components),
+            'hook_count': len(hooks),
+        }
+
+
+# ── Import Chain Tracer ───────────────────────────────────────────────────────
+
+class ImportChainTracer:
+    MAX_DEPTH = 5
+
+    def trace(self, routes: list[dict]) -> list[dict]:
+        chains = []
+        for route in routes[:30]:  # limit to first 30 routes
+            f = PROJECT_ROOT / route.get('file', '')
+            if not f.exists() or f.suffix != '.py':
+                continue
+            chain = self._trace_file(f, depth=0)
+            if len(chain) > 1:
+                chains.append({
+                    'route': f"{route.get('method','?')} {route.get('path','?')}",
+                    'chain': ' → '.join(chain),
+                    'file': route.get('file'),
+                })
+        return chains
+
+    def _trace_file(self, f: Path, depth: int) -> list[str]:
+        if depth > self.MAX_DEPTH or not f.exists():
+            return []
+        result = [str(f.relative_to(PROJECT_ROOT))]
+        try:
+            source = f.read_text(encoding='utf-8', errors='replace')
+            tree = ast.parse(source)
+            for node in ast.walk(tree):
+                if isinstance(node, (ast.Import, ast.ImportFrom)):
+                    if isinstance(node, ast.ImportFrom) and node.module:
+                        # Attempt to resolve local import
+                        parts = node.module.split('.')
+                        candidate = PROJECT_ROOT / Path(*parts).with_suffix('.py')
+                        if candidate.exists() and depth < self.MAX_DEPTH:
+                            sub = self._trace_file(candidate, depth + 1)
+                            if sub:
+                                result.extend(sub[1:])
+                                break
+        except Exception:
+            pass
+        return result
+
+
+# ── Vocabulary Builder ────────────────────────────────────────────────────────
+
+class VocabularyBuilder:
+    SKIP_WORDS = {'index', 'main', 'base', 'common', 'utils', 'helpers', 'types',
+                  'constants', 'config', 'lib', 'api', 'app', 'src', 'test', 'tests'}
+
+    def build(
+        self,
+        routes: list[dict],
+        models: list[dict],
+        schemas: list[dict],
+        features: list[dict],
+        stack: dict,
+    ) -> list[dict]:
+        vocab: dict[str, dict] = {}
+
+        # From features
+        for feat in features:
+            name = feat['name']
+            if name.lower() in self.SKIP_WORDS:
+                continue
+            aliases = self._name_to_aliases(name)
+            for alias in aliases:
+                self._add(vocab, alias, 'feature', feat.get('path', ''), f"{feat['component_count']} components")
+
+        # From models
+        for model in models:
+            aliases = self._name_to_aliases(model['name'])
+            for alias in aliases:
+                self._add(vocab, alias, 'model', model.get('file', ''), f"table/model: {model['name']}")
+
+        # From routes (group by path prefix)
+        route_groups: dict[str, list] = {}
+        for route in routes:
+            prefix = route.get('path', '/').split('/')[1] if '/' in route.get('path', '/') else route.get('path', '')
+            if prefix and prefix not in self.SKIP_WORDS:
+                route_groups.setdefault(prefix, []).append(route)
+        for prefix, group in route_groups.items():
+            aliases = self._name_to_aliases(prefix)
+            file_ex = group[0].get('file', '')
+            for alias in aliases:
+                self._add(vocab, alias, 'api', file_ex, f"{len(group)} routes")
+
+        # Merge learned vocabulary
+        learned = self._load_learned()
+        for alias, data in learned.items():
+            if alias not in vocab and data.get('score', 0) >= 5:
+                vocab[alias] = {
+                    'alias': alias,
+                    'type': 'learned',
+                    'location': ', '.join(data.get('targets', [])),
+                    'notes': f"learned from session (score: {data['score']:.1f})",
+                }
+
+        return list(vocab.values())
+
+    def _name_to_aliases(self, name: str) -> list[str]:
+        """Generate human aliases from a code name."""
+        # camelCase / PascalCase → words
+        words = re.sub(r'([A-Z])', r' \1', name).lower().split()
+        words = [w for w in words if w and w not in self.SKIP_WORDS]
+        if not words:
+            return []
+        phrase = ' '.join(words)
+        aliases = [phrase, name.lower()]
+        # e.g. 'deal-pipeline' → 'deal pipeline'
+        if '-' in name or '_' in name:
+            clean = re.sub(r'[-_]', ' ', name).lower()
+            aliases.append(clean)
+        return list(dict.fromkeys(aliases))  # dedupe, preserve order
+
+    def _add(self, vocab: dict, alias: str, type_: str, location: str, notes: str) -> None:
+        if alias and alias not in vocab:
+            vocab[alias] = {'alias': alias, 'type': type_, 'location': location, 'notes': notes}
+
+    def _load_learned(self) -> dict:
+        if LEARNED_VOC.exists():
+            try:
+                return json.loads(LEARNED_VOC.read_text())
+            except Exception:
+                pass
+        return {}
+
+
+# ── Tools & Commands Scanner ──────────────────────────────────────────────────
+
+class ToolsScanner:
+    def scan(self) -> list[dict]:
+        tools = []
+
+        # npm scripts
+        pkg = PROJECT_ROOT / 'package.json'
+        if pkg.exists():
+            try:
+                data = json.loads(pkg.read_text())
+                for name, cmd in (data.get('scripts') or {}).items():
+                    tools.append({'name': name, 'command': f'npm run {name}', 'description': cmd, 'source': 'package.json'})
+            except Exception:
+                pass
+
+        # Makefile targets
+        makefile = PROJECT_ROOT / 'Makefile'
+        if makefile.exists():
+            for line in makefile.read_text(encoding='utf-8', errors='replace').splitlines():
+                m = re.match(r'^([a-zA-Z][a-zA-Z0-9_-]+)\s*:', line)
+                if m and not m.group(1).startswith('.'):
+                    tools.append({'name': m.group(1), 'command': f'make {m.group(1)}', 'description': '', 'source': 'Makefile'})
+
+        # Poetry / pip scripts
+        for cfg in ['pyproject.toml', 'setup.cfg']:
+            path = PROJECT_ROOT / cfg
+            if path.exists():
+                text = path.read_text(encoding='utf-8', errors='replace')
+                for m in re.finditer(r'^\[tool\.poetry\.scripts\]\s*\n((?:\w.*\n)*)', text, re.MULTILINE):
+                    for line in m.group(1).splitlines():
+                        if '=' in line:
+                            name = line.split('=')[0].strip()
+                            tools.append({'name': name, 'command': name, 'description': '', 'source': cfg})
+
+        # Shell scripts at root
+        for f in PROJECT_ROOT.glob('*.sh'):
+            tools.append({'name': f.name, 'command': f'bash {f.name}', 'description': '', 'source': 'shell'})
+
+        # .claude/skills
+        skills_dir = PROJECT_ROOT / '.claude' / 'skills'
+        if skills_dir.is_dir():
+            for skill_dir in skills_dir.iterdir():
+                if (skill_dir / 'SKILL.md').exists():
+                    tools.append({'name': f'/{skill_dir.name}', 'command': f'/{skill_dir.name}', 'description': 'Claude skill', 'source': 'skills'})
+
+        return tools
+
+
+# ── Auth Config Scanner ───────────────────────────────────────────────────────
+
+class AuthScanner:
+    def scan(self) -> dict:
+        info: dict[str, Any] = {'provider': 'unknown', 'files': [], 'patterns': []}
+
+        patterns_map = {
+            'supabase': ['supabase', 'createClient', 'auth.signIn'],
+            'next-auth': ['NextAuth', 'getSession', 'useSession', 'SessionProvider'],
+            'clerk': ['ClerkProvider', 'useUser', '@clerk'],
+            'auth0': ['Auth0Provider', 'useAuth0', '@auth0'],
+            'jwt': ['jwt.sign', 'jwt.verify', 'create_access_token', 'decode_token'],
+            'passport': ['passport.use', 'passport.authenticate'],
+            'firebase': ['initializeApp', 'getAuth', 'signInWithEmailAndPassword'],
+        }
+
+        for f in PROJECT_ROOT.rglob('*'):
+            if any(p in IGNORE_DIRS for p in f.parts):
+                continue
+            if f.suffix not in ('.py', '.ts', '.tsx', '.js', '.jsx'):
+                continue
+            try:
+                source = f.read_text(encoding='utf-8', errors='replace')
+            except Exception:
+                continue
+            for provider, patterns in patterns_map.items():
+                if any(p in source for p in patterns):
+                    info['provider'] = provider
+                    rel = str(f.relative_to(PROJECT_ROOT))
+                    if rel not in info['files']:
+                        info['files'].append(rel)
+                    info['patterns'].extend([p for p in patterns if p in source and p not in info['patterns']])
+
+        return info
+
+
+# ── Reverse Proxy Scanner ─────────────────────────────────────────────────────
+
+class ReverseProxyScanner:
+    def scan(self) -> list[dict]:
+        rules = []
+        # nginx
+        for f in PROJECT_ROOT.rglob('*.conf'):
+            if any(p in IGNORE_DIRS for p in f.parts):
+                continue
+            try:
+                source = f.read_text(encoding='utf-8', errors='replace')
+                for m in re.finditer(r'location\s+([^\s{]+)\s*\{[^}]*proxy_pass\s+([^;]+);', source, re.DOTALL):
+                    rules.append({'type': 'nginx', 'location': m.group(1).strip(), 'upstream': m.group(2).strip(), 'file': str(f.relative_to(PROJECT_ROOT))})
+            except Exception:
+                pass
+        # Caddyfile
+        caddyfile = PROJECT_ROOT / 'Caddyfile'
+        if caddyfile.exists():
+            try:
+                source = caddyfile.read_text(encoding='utf-8', errors='replace')
+                for m in re.finditer(r'reverse_proxy\s+([^\n]+)', source):
+                    rules.append({'type': 'caddy', 'location': '*', 'upstream': m.group(1).strip(), 'file': 'Caddyfile'})
+            except Exception:
+                pass
+        return rules
+
+
+# ── Dead Code Detector ────────────────────────────────────────────────────────
+
+class DeadCodeDetector:
+    def detect(self, vocab: list[dict]) -> list[dict]:
+        candidates = []
+        cutoff_date = '--since=6 months ago'
+
+        low_score_files = set()
+        if LEARNED_VOC.exists():
+            try:
+                learned = json.loads(LEARNED_VOC.read_text())
+                for alias, data in learned.items():
+                    if data.get('score', 999) < 2:
+                        low_score_files.update(data.get('targets', []))
+            except Exception:
+                pass
+
+        for filepath_str in low_score_files:
+            path = PROJECT_ROOT / filepath_str
+            if not path.exists():
+                continue
+            try:
+                result = subprocess.run(
+                    ['git', '-C', str(PROJECT_ROOT), 'log', cutoff_date, '--', filepath_str],
+                    capture_output=True, text=True, timeout=5
+                )
+                if not result.stdout.strip():
+                    candidates.append({
+                        'file': filepath_str,
+                        'reason': 'No commits in 6 months + low vocabulary score',
+                    })
+            except Exception:
+                pass
+
+        return candidates
+
+
+# ╔══════════════════════════════════════════════════════════════════════════╗
+# ║  SECTION WRITERS                                                         ║
+# ╚══════════════════════════════════════════════════════════════════════════╝
+
+def write_section(num: str, name: str, content: str) -> Path:
+    filename = f"{num}-{name}.md"
+    path = SECTIONS_DIR / filename
+    path.write_text(content, encoding='utf-8')
+    return path
+
+def section_size_kb(path: Path) -> float:
+    return path.stat().st_size / 1024 if path.exists() else 0.0
+
+
+def build_vocabulary_section(vocab: list[dict]) -> str:
+    lines = [
+        "# Section 01 — Vocabulary Translation Layer\n",
+        "> Maps human language to exact code locations. Auto-generated.\n\n",
+        "| Alias | Type | Location | Notes |",
+        "|-------|------|----------|-------|",
+    ]
+    for v in sorted(vocab, key=lambda x: x.get('alias', '')):
+        alias    = v.get('alias', '').replace('|', '\\|')
+        type_    = v.get('type', '').replace('|', '\\|')
+        location = v.get('location', '').replace('|', '\\|')
+        notes    = v.get('notes', '').replace('|', '\\|')
+        lines.append(f"| {alias} | {type_} | {location} | {notes} |")
+    if not vocab:
+        lines.append("| _(no vocabulary generated yet — add source code to populate)_ | | | |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_topology_section(services: list[dict]) -> str:
+    lines = [
+        "# Section 02 — Service Topology\n",
+        "> Docker services, ports, and dependencies.\n\n",
+    ]
+    if not services:
+        lines.append("_No docker-compose services detected._\n")
+        return '\n'.join(lines)
+    lines += ["| Service | Image | Ports | Depends On |",
+              "|---------|-------|-------|------------|"]
+    for svc in services:
+        ports = ', '.join(str(p) for p in svc.get('ports', []))
+        deps  = ', '.join(svc.get('depends_on', []) if isinstance(svc.get('depends_on'), list)
+                         else list(svc.get('depends_on', {}).keys()))
+        img   = str(svc.get('image', '?'))[:50]
+        lines.append(f"| {svc['name']} | {img} | {ports} | {deps} |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_environment_section(env_entries: list[dict]) -> str:
+    lines = [
+        "# Section 03 — Environment Variables\n",
+        "> Key names and non-sensitive values only. Secrets are redacted.\n\n",
+    ]
+    if not env_entries:
+        lines.append("_No .env files found._\n")
+        return '\n'.join(lines)
+    lines += ["| Key | Value | Source |",
+              "|-----|-------|--------|"]
+    for e in env_entries:
+        val = '[SECRET]' if e.get('is_secret') else str(e.get('value', ''))[:60]
+        lines.append(f"| `{e['key']}` | `{val}` | {e.get('source', '')} |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_routes_section(routes: list[dict]) -> str:
+    lines = [
+        "# Section 04 — API Routes\n\n",
+        "| Method | Path | File | Line |",
+        "|--------|------|------|------|",
+    ]
+    for r in sorted(routes, key=lambda x: (x.get('path',''), x.get('method',''))):
+        lines.append(f"| `{r.get('method','?')}` | `{r.get('path','?')}` | {r.get('file','')} | {r.get('line','')} |")
+    if not routes:
+        lines.append("| _(no routes detected yet)_ | | | |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_models_section(models: list[dict]) -> str:
+    lines = ["# Section 05 — Data Models\n\n"]
+    if not models:
+        lines.append("_No data models detected yet._\n")
+        return '\n'.join(lines)
+    for model in models:
+        lines.append(f"## {model['name']}")
+        lines.append(f"- **File**: `{model.get('file','?')}`")
+        cols = model.get('columns', [])
+        if cols:
+            lines.append("- **Fields**:")
+            for c in cols[:20]:
+                lines.append(f"  - `{c.get('name','?')}` ({c.get('type','?')})")
+        lines.append('')
+    return '\n'.join(lines)
+
+
+def build_schemas_section(schemas: list[dict]) -> str:
+    lines = ["# Section 06 — Schemas / DTOs\n\n"]
+    if not schemas:
+        lines.append("_No schemas detected yet._\n")
+        return '\n'.join(lines)
+    for s in schemas:
+        lines.append(f"## {s['name']}")
+        lines.append(f"- **File**: `{s.get('file','?')}`")
+        fields = s.get('fields', [])
+        if fields:
+            lines.append(f"- **Fields**: {', '.join(f'`{f}`' for f in fields[:15])}")
+        lines.append('')
+    return '\n'.join(lines)
+
+
+def build_services_section(routes: list[dict], models: list[dict]) -> str:
+    lines = ["# Section 07 — Services\n\n",
+             "_Service discovery is inferred from directory structure and imports._\n\n"]
+    # Collect unique directories containing routes or models
+    dirs: dict[str, int] = {}
+    for item in routes + models:
+        f = item.get('file', '')
+        d = str(Path(f).parent) if f else ''
+        if d and d != '.':
+            dirs[d] = dirs.get(d, 0) + 1
+    if dirs:
+        lines += ["| Directory | Items |", "|-----------|-------|"]
+        for d, count in sorted(dirs.items(), key=lambda x: -x[1]):
+            lines.append(f"| `{d}` | {count} |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_background_jobs_section() -> str:
+    lines = ["# Section 08 — Background Jobs\n\n"]
+    jobs = []
+    # Celery
+    for f in PROJECT_ROOT.rglob('*.py'):
+        if any(p in IGNORE_DIRS for p in f.parts):
+            continue
+        try:
+            source = f.read_text(encoding='utf-8', errors='replace')
+            for m in re.finditer(r'@(?:app|celery)\.task|@shared_task', source):
+                # Find function after decorator
+                fn_m = re.search(r'def\s+(\w+)\s*\(', source[m.start():m.start()+200])
+                if fn_m:
+                    line = source[:m.start()].count('\n') + 1
+                    jobs.append({'name': fn_m.group(1), 'type': 'celery', 'file': str(f.relative_to(PROJECT_ROOT)), 'line': line})
+        except Exception:
+            pass
+    # Cron / APScheduler
+    for f in PROJECT_ROOT.rglob('*.py'):
+        if any(p in IGNORE_DIRS for p in f.parts):
+            continue
+        try:
+            source = f.read_text(encoding='utf-8', errors='replace')
+            for m in re.finditer(r'@scheduler\.scheduled_job|scheduler\.add_job|@cron', source):
+                fn_m = re.search(r'def\s+(\w+)\s*\(', source[m.start():m.start()+200])
+                if fn_m:
+                    line = source[:m.start()].count('\n') + 1
+                    jobs.append({'name': fn_m.group(1), 'type': 'scheduler', 'file': str(f.relative_to(PROJECT_ROOT)), 'line': line})
+        except Exception:
+            pass
+
+    if not jobs:
+        lines.append("_No background jobs detected._\n")
+    else:
+        lines += ["| Job | Type | File | Line |", "|-----|------|------|------|"]
+        for j in jobs:
+            lines.append(f"| `{j['name']}` | {j['type']} | {j['file']} | {j.get('line','')} |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_frontend_section(features: list[dict]) -> str:
+    lines = ["# Section 09 — Frontend Features\n\n"]
+    if not features:
+        lines.append("_No frontend feature directories detected._\n")
+        return '\n'.join(lines)
+    lines += ["| Feature | Components | Hooks | Stores | API Files |",
+              "|---------|-----------|-------|--------|-----------|"]
+    for f in features:
+        lines.append(
+            f"| `{f['path']}` | {f['component_count']} | {f['hook_count']} "
+            f"| {len(f.get('stores',[]))} | {len(f.get('api_files',[]))} |"
+        )
+    return '\n'.join(lines) + '\n'
+
+
+def build_tools_section(tools: list[dict]) -> str:
+    lines = ["# Section 10 — Tools & Commands\n\n",
+             "| Name | Command | Source |",
+             "|------|---------|--------|"]
+    for t in tools:
+        lines.append(f"| `{t['name']}` | `{t['command']}` | {t.get('source','')} |")
+    if not tools:
+        lines.append("| _(no tools detected)_ | | |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_migrations_section(migrations: list[dict]) -> str:
+    lines = ["# Section 11 — Migrations\n\n"]
+    if not migrations:
+        lines.append("_No migrations detected._\n")
+        return '\n'.join(lines)
+    lines += ["| Revision | Tables Affected | Type | File |",
+              "|----------|----------------|------|------|"]
+    for m in migrations[-30:]:  # last 30
+        tables = ', '.join(m.get('tables_affected', []))[:60]
+        lines.append(f"| `{m.get('revision','?')}` | {tables} | {m.get('type','?')} | {m.get('file','')} |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_import_chains_section(chains: list[dict]) -> str:
+    lines = ["# Section 12 — Import Chains\n\n",
+             "_Traces route → service → model → table for key endpoints._\n\n"]
+    if not chains:
+        lines.append("_No import chains traced (requires Python source files with routes)._\n")
+        return '\n'.join(lines)
+    for c in chains:
+        lines.append(f"**{c['route']}**")
+        lines.append(f"```\n{c['chain']}\n```\n")
+    return '\n'.join(lines)
+
+
+def build_frontend_backend_section(routes: list[dict], features: list[dict]) -> str:
+    lines = ["# Section 13 — Frontend → Backend Map\n\n",
+             "_Maps frontend API service calls to backend route paths._\n\n"]
+    mappings = []
+    api_calls: list[tuple[str,str]] = []
+
+    for feat in features:
+        for api_file in feat.get('api_files', []):
+            full = PROJECT_ROOT / feat['path'] / api_file
+            if full.exists():
+                try:
+                    source = full.read_text(encoding='utf-8', errors='replace')
+                    for m in re.finditer(r'[\'"`](/api/[^\'"` \n]+)', source):
+                        api_calls.append((m.group(1), str(full.relative_to(PROJECT_ROOT))))
+                except Exception:
+                    pass
+
+    route_paths = {r.get('path',''): r for r in routes}
+    for call, fe_file in api_calls:
+        if call in route_paths:
+            be = route_paths[call]
+            mappings.append({'frontend': fe_file, 'url': call, 'backend': be.get('file','?')})
+
+    if not mappings:
+        lines.append("_No frontend→backend mappings detected yet._\n")
+    else:
+        lines += ["| Frontend File | URL | Backend File |",
+                  "|--------------|-----|--------------|"]
+        for m in mappings:
+            lines.append(f"| {m['frontend']} | `{m['url']}` | {m['backend']} |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_proxy_section(proxy_rules: list[dict]) -> str:
+    lines = ["# Section 14 — Reverse Proxy\n\n"]
+    if not proxy_rules:
+        lines.append("_No reverse proxy configuration detected._\n")
+        return '\n'.join(lines)
+    lines += ["| Type | Location | Upstream | File |",
+              "|------|----------|----------|------|"]
+    for r in proxy_rules:
+        lines.append(f"| {r['type']} | `{r['location']}` | `{r['upstream']}` | {r['file']} |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_auth_section(auth_info: dict) -> str:
+    lines = ["# Section 15 — Auth Configuration\n\n",
+             f"**Provider**: {auth_info.get('provider','unknown')}\n\n"]
+    files = auth_info.get('files', [])
+    if files:
+        lines.append("**Auth files:**")
+        for f in files[:10]:
+            lines.append(f"- `{f}`")
+    patterns = auth_info.get('patterns', [])
+    if patterns:
+        lines.append(f"\n**Detected patterns**: {', '.join(f'`{p}`' for p in patterns[:10])}")
+    return '\n'.join(lines) + '\n'
+
+
+def build_infra_section(stack: dict, services: list[dict]) -> str:
+    lines = ["# Section 16 — Infrastructure Profile\n\n",
+             f"| Property | Value |",
+             "|----------|-------|",
+             f"| **Language** | {stack.get('language','?')} |",
+             f"| **Framework** | {stack.get('framework','?')} |",
+             f"| **Database** | {stack.get('database','?')} |",
+             f"| **ORM** | {stack.get('orm','?')} |",
+             f"| **Auth** | {stack.get('auth','?')} |",
+             f"| **Package Manager** | {stack.get('package_manager','?')} |",
+             f"| **Infrastructure** | {stack.get('infrastructure','?')} |",
+             f"| **Services** | {len(services)} docker services |",
+             ""]
+    return '\n'.join(lines)
+
+
+def build_learned_vocab_section() -> str:
+    lines = ["# Section 17 — Learned Vocabulary\n\n",
+             "_Aliases mined from Claude Code session history. Score = frequency × recency._\n\n"]
+    learned = {}
+    if LEARNED_VOC.exists():
+        try:
+            learned = json.loads(LEARNED_VOC.read_text())
+        except Exception:
+            pass
+    if not learned:
+        lines.append("_No session-mined vocabulary yet. Accumulates over time._\n")
+        return '\n'.join(lines)
+    lines += ["| Alias | Score | Targets | Last Seen |",
+              "|-------|-------|---------|-----------|"]
+    for alias, data in sorted(learned.items(), key=lambda x: -x[1].get('score', 0)):
+        if data.get('score', 0) >= 2:
+            targets = ', '.join(data.get('targets', []))[:60]
+            lines.append(f"| {alias} | {data.get('score',0):.1f} | {targets} | {data.get('last_seen','?')} |")
+    return '\n'.join(lines) + '\n'
+
+
+def build_dead_code_section(candidates: list[dict]) -> str:
+    lines = ["# Section 18 — Dead Code Candidates\n\n",
+             "> Files flagged for human review: no recent commits AND low vocabulary score.\n",
+             "> **Do not auto-delete.** Review before removing.\n\n"]
+    if not candidates:
+        lines.append("_No dead code candidates detected._\n")
+        return '\n'.join(lines)
+    for c in candidates:
+        lines.append(f"- `{c['file']}` — {c.get('reason','')}")
+    return '\n'.join(lines) + '\n'
+
+
+def build_doc_pointers_section() -> str:
+    lines = ["# Section 19 — Documentation Pointers\n\n"]
+    docs = []
+    doc_dirs = ['docs', 'doc', 'documentation', 'wiki', '.docs']
+    doc_exts = {'.md', '.rst', '.txt', '.adoc'}
+
+    for doc_dir in doc_dirs:
+        d = PROJECT_ROOT / doc_dir
+        if d.is_dir():
+            for f in sorted(d.rglob('*')):
+                if f.is_file() and f.suffix in doc_exts:
+                    docs.append(str(f.relative_to(PROJECT_ROOT)))
+
+    # Root-level docs
+    for f in PROJECT_ROOT.glob('*.md'):
+        docs.append(str(f.relative_to(PROJECT_ROOT)))
+
+    if not docs:
+        lines.append("_No documentation files found._\n")
+    else:
+        for d in docs[:30]:
+            lines.append(f"- [`{d}`]({d})")
+    return '\n'.join(lines) + '\n'
+
+
+# ╔══════════════════════════════════════════════════════════════════════════╗
+# ║  PROJECT MAP TOC                                                         ║
+# ╚══════════════════════════════════════════════════════════════════════════╝
+
+def build_project_map(
+    stack: dict,
+    routes: list[dict],
+    models: list[dict],
+    schemas: list[dict],
+    features: list[dict],
+    migrations: list[dict],
+    services: list[dict],
+    vocab: list[dict],
+    section_files: list[tuple[str, Path]],
+) -> str:
+    now = datetime.now().strftime('%Y-%m-%d %H:%M')
+    name = stack.get('name', PROJECT_ROOT.name)
+    slug = stack.get('slug', name)
+
+    lines = [
+        f"# {name} — Project Map\n",
+        f"> Auto-generated by `generate.py` on {now}. Do not edit manually.\n\n",
+        f"## Stats\n",
+        f"| Metric | Count |",
+        f"|--------|-------|",
+        f"| API Routes | {len(routes)} |",
+        f"| Data Models | {len(models)} |",
+        f"| Schemas/DTOs | {len(schemas)} |",
+        f"| Frontend Features | {len(features)} |",
+        f"| Migrations | {len(migrations)} |",
+        f"| Docker Services | {len(services)} |",
+        f"| Vocabulary Entries | {len(vocab)} |",
+        f"| Stack | {stack.get('language','?')} / {stack.get('framework','?')} |",
+        "",
+        "## Section Index\n",
+        "| # | Section | Size | When to Read |",
+        "|---|---------|------|--------------|",
+    ]
+
+    WHEN_TO_READ = {
+        '01': 'Any task — start here if you don\'t know where the code lives',
+        '02': 'Debugging connectivity, adding a service, understanding ports',
+        '03': 'Environment setup, missing vars, config issues',
+        '04': 'Adding/editing API endpoints, checking what routes exist',
+        '05': 'Changing database schema, adding fields, understanding relations',
+        '06': 'Adding DTOs, changing request/response shapes',
+        '07': 'Adding service logic, understanding service boundaries',
+        '08': 'Working with background jobs, queues, scheduled tasks',
+        '09': 'Frontend feature work, understanding UI structure',
+        '10': 'Available commands, scripts, developer tooling',
+        '11': 'Database migrations, schema history',
+        '12': 'Tracing data flow from HTTP request to DB',
+        '13': 'Understanding which frontend calls which backend endpoint',
+        '14': 'Proxy routing, nginx/caddy config',
+        '15': 'Auth flow, sessions, permissions',
+        '16': 'Infrastructure overview, tech stack summary',
+        '17': 'Vocabulary learned from past sessions',
+        '18': 'Dead code review',
+        '19': 'Finding documentation, READMEs, wikis',
+    }
+
+    for name_part, path in section_files:
+        num = name_part.split('-')[0]
+        display = name_part.replace('-', ' ').title()
+        size_kb = section_size_kb(path)
+        when = WHEN_TO_READ.get(num, '')
+        lines.append(f"| [{num}](sections/{path.name}) | {display} | {size_kb:.1f} KB | {when} |")
+
+    lines += [
+        "",
+        "## Quick Routing\n",
+        "| Task | Read Sections |",
+        "|------|--------------|",
+        "| Feature / UX work | 01 → 09 → 04 |",
+        "| Add model or field | 05 → 06 → 12 |",
+        "| Troubleshoot error | 02 → 03 → 14 |",
+        "| Infrastructure / scaling | 16 → 02 |",
+        "| Auth / security | 15 → 19 |",
+        "| What tools exist | 10 |",
+        "| Background jobs | 08 |",
+        "| Migration history | 11 |",
+        "",
+        f"## Regenerate\n",
+        "```bash",
+        "python .claude/project-map/generate.py          # skip if unchanged",
+        "python .claude/project-map/generate.py --force  # always regenerate",
+        "```",
+    ]
+    return '\n'.join(lines) + '\n'
+
+
+# ╔══════════════════════════════════════════════════════════════════════════╗
+# ║  MAIN                                                                    ║
+# ╚══════════════════════════════════════════════════════════════════════════╝
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description='Babel Fish — generate project map')
+    parser.add_argument('--force', action='store_true', help='Force regeneration even if checksums match')
+    parser.add_argument('--project-root', type=Path, default=None, help='Override project root')
+    parser.add_argument('--stack-json', type=Path, default=None, help='Path to stack.json from detect-stack.sh')
+    args = parser.parse_args()
+
+    global PROJECT_ROOT
+    if args.project_root:
+        PROJECT_ROOT = args.project_root.resolve()
+
+    print(f"[generate] Project root: {PROJECT_ROOT}")
+
+    # 1. Checksum check
+    watched = collect_watched_files()
+    checksum = compute_checksum(watched)
+
+    if not args.force and is_unchanged(checksum):
+        print("[generate] ✓ No changes detected — skipping regeneration (use --force to override)")
+        sys.exit(0)
+
+    # 2. Load stack
+    stack = load_stack(args.stack_json)
+    print(f"[generate] Stack: {stack['language']} / {stack['framework']}")
+
+    # 3. Collect all relevant files
+    py_files  = [f for f in watched if f.suffix == '.py']
+    ts_files  = [f for f in watched if f.suffix in ('.ts','.tsx','.js','.jsx')]
+    go_files  = [f for f in watched if f.suffix == '.go']
+
+    # 4. Parse
+    print("[generate] Parsing routes...")
+    routes: list[dict] = []
+    if stack['language'] in ('python', 'unknown'):
+        routes.extend(PythonRouteParser().parse(py_files))
+    if stack['language'] in ('typescript', 'javascript', 'unknown'):
+        routes.extend(TSRouteParser().parse(ts_files))
+    if stack['language'] in ('go', 'unknown'):
+        routes.extend(GoRouteParser().parse(go_files))
+
+    print("[generate] Parsing models...")
+    models: list[dict] = []
+    if stack['language'] in ('python', 'unknown'):
+        models.extend(PythonModelParser().parse(py_files))
+    if stack['language'] in ('typescript', 'javascript', 'unknown'):
+        models.extend(TSModelParser().parse(ts_files + [f for f in watched if f.suffix == '.prisma']))
+    if stack['language'] in ('go', 'unknown'):
+        models.extend(GoModelParser().parse(go_files))
+
+    print("[generate] Parsing schemas...")
+    schemas = PythonSchemaParser().parse(py_files)
+
+    print("[generate] Scanning environment...")
+    services = DockerComposeParser().parse()
+    env_entries = EnvParser().parse()
+    migrations = MigrationParser().parse()
+    features = FrontendScanner().scan()
+    tools = ToolsScanner().scan()
+    auth_info = AuthScanner().scan()
+    proxy_rules = ReverseProxyScanner().scan()
+
+    print("[generate] Building vocabulary...")
+    vocab = VocabularyBuilder().build(routes, models, schemas, features, stack)
+
+    print("[generate] Tracing import chains...")
+    chains = ImportChainTracer().trace(routes)
+
+    print("[generate] Detecting dead code candidates...")
+    dead_code = DeadCodeDetector().detect(vocab)
+
+    # 5. Write sections
+    print("[generate] Writing sections...")
+    section_files: list[tuple[str, Path]] = [
+        ('01-vocabulary',          write_section('01', 'vocabulary',          build_vocabulary_section(vocab))),
+        ('02-service-topology',    write_section('02', 'service-topology',    build_topology_section(services))),
+        ('03-environment',         write_section('03', 'environment',         build_environment_section(env_entries))),
+        ('04-api-routes',          write_section('04', 'api-routes',          build_routes_section(routes))),
+        ('05-data-models',         write_section('05', 'data-models',         build_models_section(models))),
+        ('06-schemas',             write_section('06', 'schemas',             build_schemas_section(schemas))),
+        ('07-services',            write_section('07', 'services',            build_services_section(routes, models))),
+        ('08-background-jobs',     write_section('08', 'background-jobs',     build_background_jobs_section())),
+        ('09-frontend-features',   write_section('09', 'frontend-features',   build_frontend_section(features))),
+        ('10-tools-commands',      write_section('10', 'tools-commands',      build_tools_section(tools))),
+        ('11-migrations',          write_section('11', 'migrations',          build_migrations_section(migrations))),
+        ('12-import-chains',       write_section('12', 'import-chains',       build_import_chains_section(chains))),
+        ('13-frontend-backend-map',write_section('13', 'frontend-backend-map',build_frontend_backend_section(routes, features))),
+        ('14-reverse-proxy',       write_section('14', 'reverse-proxy',       build_proxy_section(proxy_rules))),
+        ('15-auth-config',         write_section('15', 'auth-config',         build_auth_section(auth_info))),
+        ('16-infra-profile',       write_section('16', 'infra-profile',       build_infra_section(stack, services))),
+        ('17-learned-vocabulary',  write_section('17', 'learned-vocabulary',  build_learned_vocab_section())),
+        ('18-dead-code',           write_section('18', 'dead-code',           build_dead_code_section(dead_code))),
+        ('19-doc-pointers',        write_section('19', 'doc-pointers',        build_doc_pointers_section())),
+    ]
+
+    # 6. Write PROJECT_MAP.md
+    print("[generate] Writing PROJECT_MAP.md...")
+    project_map = build_project_map(stack, routes, models, schemas, features, migrations, services, vocab, section_files)
+    (MAP_DIR / 'PROJECT_MAP.md').write_text(project_map, encoding='utf-8')
+
+    # 7. Update checksums
+    save_checksums({'input_hash': checksum, 'generated_at': datetime.now().isoformat(), 'route_count': len(routes), 'model_count': len(models)})
+
+    # 8. Ensure learned-vocabulary.json exists
+    if not LEARNED_VOC.exists():
+        LEARNED_VOC.write_text('{}', encoding='utf-8')
+
+    total_kb = sum(section_size_kb(p) for _, p in section_files)
+    print(f"[generate] ✓ Done — {len(routes)} routes, {len(models)} models, {len(vocab)} vocab entries, {total_kb:.1f} KB total")
+
+
+if __name__ == '__main__':
+    main()