fix: add automatic temp folder cleanup at Maestro startup

Problem:
When AutoForge runs agents that use Playwright for browser testing or
mongodb-memory-server for database tests, temporary files accumulate in
the system temp folder (%TEMP% on Windows, /tmp on Linux/macOS). These
files are never cleaned up automatically and can consume hundreds of GB
over time.

Affected temp items:
- playwright_firefoxdev_profile-* (browser profiles)
- playwright-artifacts-* (test artifacts)
- playwright-transform-cache
- mongodb-memory-server* (MongoDB binaries)
- ng-* (Angular CLI temp)
- scoped_dir* (Chrome/Chromium temp)
- .78912*.node (Node.js native module cache, ~7MB each)
- claude-*-cwd (Claude CLI working directory files)
- mat-debug-*.log (Material/Angular debug logs)

Solution:
- New temp_cleanup.py module with cleanup_stale_temp() function
- Called at Maestro (orchestrator) startup in autonomous_agent_demo.py
- Only deletes files/folders older than 1 hour (safe for running processes)
- Runs every time the Play button is clicked or agent auto-restarts
- Reports cleanup stats: dirs deleted, files deleted, MB freed

Why cleanup at Maestro startup:
- Reliable hook point (runs on every agent start, including auto-restart
  after rate limits which happens every ~5 hours)
- No need for background timers or scheduled tasks
- Cleanup happens before new temp files are created

Testing:
- Tested on Windows with 958 items in temp folder
- Successfully cleaned 45 dirs, 758 files, freed 415 MB
- Files younger than 1 hour correctly preserved

Closes #155

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

autonomous_agent_demo.py

@@ -263,6 +263,17 @@ def main() -> None:
             )
         else:
             # Entry point mode - always use unified orchestrator
+            # Clean up stale temp files before starting (prevents temp folder bloat)
+            from temp_cleanup import cleanup_stale_temp
+            cleanup_stats = cleanup_stale_temp()
+            if cleanup_stats["dirs_deleted"] > 0 or cleanup_stats["files_deleted"] > 0:
+                mb_freed = cleanup_stats["bytes_freed"] / (1024 * 1024)
+                print(
+                    f"[CLEANUP] Removed {cleanup_stats['dirs_deleted']} dirs, "
+                    f"{cleanup_stats['files_deleted']} files ({mb_freed:.1f} MB freed)",
+                    flush=True,
+                )
+
             from parallel_orchestrator import run_parallel_orchestrator
 
             # Clamp concurrency to valid range (1-5)

temp_cleanup.py

@@ -0,0 +1,148 @@
+"""
+Temp Cleanup Module
+===================
+
+Cleans up stale temporary files and directories created by AutoForge agents,
+Playwright, Node.js, and other development tools.
+
+Called at Maestro (orchestrator) startup to prevent temp folder bloat.
+
+Why this exists:
+- Playwright creates browser profiles and artifacts in %TEMP%
+- Node.js creates .node cache files (~7MB each, can accumulate to GBs)
+- MongoDB Memory Server downloads binaries to temp
+- These are never cleaned up automatically
+
+When cleanup runs:
+- At Maestro startup (when you click Play or auto-restart after rate limits)
+- Only files/folders older than 1 hour are deleted (safe for running processes)
+"""
+
+import logging
+import shutil
+import tempfile
+import time
+from pathlib import Path
+
+logger = logging.getLogger(__name__)
+
+# Max age in seconds before a temp item is considered stale (1 hour)
+MAX_AGE_SECONDS = 3600
+
+# Directory patterns to clean up (glob patterns)
+DIR_PATTERNS = [
+    "playwright_firefoxdev_profile-*",  # Playwright Firefox profiles
+    "playwright-artifacts-*",           # Playwright test artifacts
+    "playwright-transform-cache",       # Playwright transform cache
+    "mongodb-memory-server*",           # MongoDB Memory Server binaries
+    "ng-*",                             # Angular CLI temp directories
+    "scoped_dir*",                      # Chrome/Chromium temp directories
+]
+
+# File patterns to clean up (glob patterns)
+FILE_PATTERNS = [
+    ".78912*.node",   # Node.js native module cache (major space consumer, ~7MB each)
+    "claude-*-cwd",   # Claude CLI working directory temp files
+    "mat-debug-*.log",  # Material/Angular debug logs
+]
+
+
+def cleanup_stale_temp(max_age_seconds: int = MAX_AGE_SECONDS) -> dict:
+    """
+    Clean up stale temporary files and directories.
+
+    Only deletes items older than max_age_seconds to avoid
+    interfering with currently running processes.
+
+    Args:
+        max_age_seconds: Maximum age in seconds before an item is deleted.
+                        Defaults to 1 hour (3600 seconds).
+
+    Returns:
+        Dictionary with cleanup statistics:
+        - dirs_deleted: Number of directories deleted
+        - files_deleted: Number of files deleted
+        - bytes_freed: Approximate bytes freed
+        - errors: List of error messages (for debugging, not fatal)
+    """
+    temp_dir = Path(tempfile.gettempdir())
+    cutoff_time = time.time() - max_age_seconds
+
+    stats = {
+        "dirs_deleted": 0,
+        "files_deleted": 0,
+        "bytes_freed": 0,
+        "errors": [],
+    }
+
+    # Clean up directories
+    for pattern in DIR_PATTERNS:
+        for item in temp_dir.glob(pattern):
+            if not item.is_dir():
+                continue
+            try:
+                mtime = item.stat().st_mtime
+                if mtime < cutoff_time:
+                    size = _get_dir_size(item)
+                    shutil.rmtree(item, ignore_errors=True)
+                    if not item.exists():
+                        stats["dirs_deleted"] += 1
+                        stats["bytes_freed"] += size
+                        logger.debug(f"Deleted temp directory: {item}")
+            except Exception as e:
+                stats["errors"].append(f"Failed to delete {item}: {e}")
+                logger.debug(f"Failed to delete {item}: {e}")
+
+    # Clean up files
+    for pattern in FILE_PATTERNS:
+        for item in temp_dir.glob(pattern):
+            if not item.is_file():
+                continue
+            try:
+                mtime = item.stat().st_mtime
+                if mtime < cutoff_time:
+                    size = item.stat().st_size
+                    item.unlink(missing_ok=True)
+                    if not item.exists():
+                        stats["files_deleted"] += 1
+                        stats["bytes_freed"] += size
+                        logger.debug(f"Deleted temp file: {item}")
+            except Exception as e:
+                stats["errors"].append(f"Failed to delete {item}: {e}")
+                logger.debug(f"Failed to delete {item}: {e}")
+
+    # Log summary if anything was cleaned
+    if stats["dirs_deleted"] > 0 or stats["files_deleted"] > 0:
+        mb_freed = stats["bytes_freed"] / (1024 * 1024)
+        logger.info(
+            f"Temp cleanup: {stats['dirs_deleted']} dirs, "
+            f"{stats['files_deleted']} files, {mb_freed:.1f} MB freed"
+        )
+
+    return stats
+
+
+def _get_dir_size(path: Path) -> int:
+    """Get total size of a directory in bytes."""
+    total = 0
+    try:
+        for item in path.rglob("*"):
+            if item.is_file():
+                try:
+                    total += item.stat().st_size
+                except (OSError, PermissionError):
+                    pass
+    except (OSError, PermissionError):
+        pass
+    return total
+
+
+if __name__ == "__main__":
+    # Allow running directly for testing/manual cleanup
+    logging.basicConfig(level=logging.DEBUG)
+    print("Running temp cleanup...")
+    stats = cleanup_stale_temp()
+    mb_freed = stats["bytes_freed"] / (1024 * 1024)
+    print(f"Cleanup complete: {stats['dirs_deleted']} dirs, {stats['files_deleted']} files, {mb_freed:.1f} MB freed")
+    if stats["errors"]:
+        print(f"Errors (non-fatal): {len(stats['errors'])}")