diff --git a/kbxtractor.py b/kbxtractor.py
new file mode 100644
index 0000000..28c7fa5
--- /dev/null
+++ b/kbxtractor.py
@@ -0,0 +1,346 @@
+import os
+import re
+import sys
+import json
+import mmap
+import hashlib
+import logging
+import argparse
+from typing import List, Iterable, Tuple
+from dataclasses import dataclass, field
+
+PREFIX_HEX = "4D 53 54 41 52 5F 53 45 43 55 52 45 5F 53 54 4F 52 45 5F 46 49 4C 45 5F 4D 41 47 49 43 5F 49 44"
+POSTFIX_HEX_LIST = ["00 00"]
+REQUIRE_POSTFIX_AT_EXTRACT_END = True
+PRINT_HEXVIEW = True
+EXTRACT_LEN = 228
+HEXVIEW_WIDTH = 16
+MAX_HITS = 0
+MAX_ZERO_FRACTION = 0.25
+MAX_ZERO_RUN = 16
+TAIL_START = 0x90
+MIN_TAIL_NONZERO_RATIO = 0.70
+SCRIPT_PATH = os.path.realpath(__file__)
+SCRIPT_DIR = os.path.dirname(SCRIPT_PATH)
+SCRIPT_NAME = os.path.basename(SCRIPT_PATH)
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(message)s"
+)
+log = logging.getLogger(__name__)
+
+@dataclass
+class ScanStats:
+    files_scanned: int = 0
+    candidates_found: int = 0
+    filter_rejected: int = 0
+    duplicates_skipped: int = 0
+    saved: int = 0
+
+    def report(self):
+        log.info("\n─────────────────────────────────────────")
+        log.info("  Complete")
+        log.info("─────────────────────────────────────────")
+        log.info(f"  Files scanned      : {self.files_scanned}")
+        log.info(f"  Candidates found   : {self.candidates_found}")
+        log.info(f"  Filter rejected    : {self.filter_rejected}")
+        log.info(f"  Duplicates skipped : {self.duplicates_skipped}")
+        log.info(f"  Keyboxes saved     : {self.saved}")
+        log.info("─────────────────────────────────────────")
+
+def clean_hex(s: str) -> bytes:
+    s = s.replace("0x", "").replace("0X", "")
+    s = "".join(s.split())
+    if not s:
+        return b""
+    if len(s) % 2 != 0:
+        raise ValueError("Hex string must have an even number of digits.")
+    return bytes.fromhex(s)
+
+def iter_all(mm: mmap.mmap, needle: bytes, start: int = 0) -> Iterable[int]:
+    i = start
+    while True:
+        pos = mm.find(needle, i)
+        if pos == -1:
+            return
+        yield pos
+        i = pos + 1
+
+def is_printable_ascii(b: int) -> bool:
+    return 32 <= b <= 126
+
+def hexview(data: bytes, base_offset: int = 0, width: int = 16) -> str:
+    lines = []
+    for i in range(0, len(data), width):
+        chunk = data[i:i + width]
+        hex_part = " ".join(f"{x:02X}" for x in chunk)
+        hex_part_padded = hex_part.ljust(width * 3 - 1)
+        ascii_part = "".join(chr(x) if is_printable_ascii(x) else "." for x in chunk)
+        lines.append(f"{base_offset + i:08X}  {hex_part_padded}  |{ascii_part}|")
+    return "\n".join(lines)
+
+def best_postfix_window(postfixes: List[bytes]) -> int:
+    return max((len(p) for p in postfixes if p), default=0)
+
+def find_postfix_after_extract(
+    mm: mmap.mmap,
+    postfixes: List[bytes],
+    prefix_pos: int,
+    file_size: int,
+    window: int
+) -> Tuple[int, bytes]:
+    start = prefix_pos + EXTRACT_LEN
+    if start >= file_size:
+        return -1, b""
+    end = min(file_size, start + window)
+    for pf in postfixes:
+        if not pf:
+            continue
+        if start + len(pf) > file_size:
+            continue
+        pos = mm.find(pf, start, end)
+        if pos != -1:
+            return pos, pf
+    return -1, b""
+
+def max_zero_run(b: bytes) -> int:
+    best = 0
+    cur = 0
+    for x in b:
+        if x == 0:
+            cur += 1
+            if cur > best:
+                best = cur
+        else:
+            cur = 0
+    return best
+
+def zero_fraction(b: bytes) -> float:
+    if not b:
+        return 1.0
+    return sum(1 for x in b if x == 0) / len(b)
+
+def nonzero_ratio(b: bytes) -> float:
+    if not b:
+        return 0.0
+    return sum(1 for x in b if x != 0) / len(b)
+
+def passes_filters(block: bytes, verbose: bool = False) -> bool:
+    zf = zero_fraction(block)
+    if zf > MAX_ZERO_FRACTION:
+        if verbose:
+            log.debug(f"  REJECT  zero_fraction={zf:.2%} > threshold {MAX_ZERO_FRACTION:.2%}")
+        return False
+
+    zr = max_zero_run(block)
+    if zr > MAX_ZERO_RUN:
+        if verbose:
+            log.debug(f"  REJECT  max_zero_run={zr} > threshold {MAX_ZERO_RUN}")
+        return False
+
+    tail = block[TAIL_START:] if TAIL_START < len(block) else b""
+    if tail:
+        nr = nonzero_ratio(tail)
+        if nr < MIN_TAIL_NONZERO_RATIO:
+            if verbose:
+                log.debug(f"  REJECT  tail_nonzero_ratio={nr:.2%} < threshold {MIN_TAIL_NONZERO_RATIO:.2%}")
+            return False
+
+    return True
+
+def safe_prefix_from_filename(path: str) -> str:
+    base = os.path.basename(path)
+    root, _ = os.path.splitext(base)
+    safe = []
+    for ch in root:
+        if ch.isalnum() or ch in "._-":
+            safe.append(ch)
+        else:
+            safe.append("_")
+    s = "".join(safe).strip("._-")
+    return s or "file"
+
+_KEYBOX_OUTPUT_RE = re.compile(r'_Keybox(_\d+)?\.bin$', re.IGNORECASE)
+
+def is_skipped_file(path: str) -> bool:
+    base = os.path.basename(path)
+    if base == SCRIPT_NAME:
+        return True
+    if _KEYBOX_OUTPUT_RE.search(base):
+        return True
+    return False
+
+def out_name_for_file(prefix: str, hit_idx: int, output_dir: str) -> str:
+    name = f"{prefix}_Keybox.bin" if hit_idx == 1 else f"{prefix}_Keybox_{hit_idx}.bin"
+    return os.path.join(output_dir, name)
+
+def extract_from_file(
+    path: str,
+    prefix: bytes,
+    postfixes: List[bytes],
+    output_dir: str,
+    stats: ScanStats,
+    verbose_filters: bool = False
+) -> int:
+    if not os.path.isfile(path):
+        return 0
+
+    file_size = os.path.getsize(path)
+    if file_size <= 0:
+        return 0
+
+    stats.files_scanned += 1
+    per_file_prefix = safe_prefix_from_filename(path)
+    saved = 0
+    seen = set()
+
+    window = best_postfix_window(postfixes)
+
+    with open(path, "rb") as f:
+        mm = mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ)
+        try:
+            for ppos in iter_all(mm, prefix, 0):
+                if ppos + EXTRACT_LEN > file_size:
+                    continue
+
+                qpos, _ = find_postfix_after_extract(mm, postfixes, ppos, file_size, window)
+                if qpos == -1:
+                    continue
+
+                if REQUIRE_POSTFIX_AT_EXTRACT_END and qpos != (ppos + EXTRACT_LEN):
+                    continue
+
+                stats.candidates_found += 1
+                block = mm[ppos:ppos + EXTRACT_LEN]
+
+                if not passes_filters(block, verbose=verbose_filters):
+                    stats.filter_rejected += 1
+                    continue
+
+                h = hashlib.sha256(block).digest()
+                if h in seen:
+                    stats.duplicates_skipped += 1
+                    continue
+                seen.add(h)
+
+                saved += 1
+                stats.saved += 1
+
+                out_name = out_name_for_file(per_file_prefix, saved, output_dir)
+                with open(out_name, "wb") as out:
+                    out.write(block)
+
+                log.info(f"[{os.path.basename(path)} | Hit #{saved}] -> {out_name}  prefix@0x{ppos:X}  postfix@0x{qpos:X}")
+
+                if PRINT_HEXVIEW:
+                    log.info(hexview(block, base_offset=ppos, width=HEXVIEW_WIDTH))
+                    log.info("")
+
+                if MAX_HITS and saved >= MAX_HITS:
+                    break
+        finally:
+            mm.close()
+
+    return saved
+
+def write_manifest(entries: list, output_dir: str):
+    manifest_path = os.path.join(output_dir, "manifest.json")
+    with open(manifest_path, "w") as f:
+        json.dump(entries, f, indent=2)
+    log.info(f"\nManifest written -> {manifest_path}")
+
+def main() -> int:
+    parser = argparse.ArgumentParser(
+        prog="MStarToKeybox",
+        description="Extract MStar/MediaTek Widevine L1 keyboxes from raw eMMC dumps and firmware images."
+    )
+    parser.add_argument(
+        "files",
+        nargs="*",
+        help="One or more binary files to scan."
+    )
+    parser.add_argument(
+        "--all",
+        action="store_true",
+        help="Scan all files in the current directory."
+    )
+    parser.add_argument(
+        "--output-dir", "-o",
+        default=".",
+        help="Directory to save extracted keyboxes."
+    )
+    parser.add_argument(
+        "--no-hexview",
+        action="store_true",
+        help="Remove keybox hext dump output."
+    )
+    parser.add_argument(
+        "--verbose-filters",
+        action="store_true",
+        help="Log candidate rejection reason."
+    )
+    parser.add_argument(
+        "--quiet", "-q",
+        action="store_true",
+        help="Stop all outputs."
+    )
+
+    args = parser.parse_args()
+
+    if args.quiet:
+        logging.getLogger().setLevel(logging.ERROR)
+    elif args.verbose_filters:
+        logging.getLogger().setLevel(logging.DEBUG)
+
+    global PRINT_HEXVIEW
+    if args.no_hexview:
+        PRINT_HEXVIEW = False
+
+    try:
+        prefix = clean_hex(PREFIX_HEX)
+        postfixes = [clean_hex(x) for x in POSTFIX_HEX_LIST]
+    except (ValueError, Exception) as e:
+        log.error(f"Failed to parse hex constants: {e}")
+        return 2
+
+    postfixes = [p for p in postfixes if p]
+    if not prefix or not postfixes or EXTRACT_LEN <= 0:
+        log.error("Error: prefix, postfix, or EXTRACT_LEN is missing.")
+        return 2
+
+    os.makedirs(args.output_dir, exist_ok=True)
+    stats = ScanStats()
+    manifest_entries = []
+
+    if args.all:
+        total = 0
+        for name in sorted(os.listdir(SCRIPT_DIR)):
+            path = os.path.join(SCRIPT_DIR, name)
+            if not os.path.isfile(path):
+                continue
+            if is_skipped_file(path):
+                continue
+            total += extract_from_file(path, prefix, postfixes, args.output_dir, stats, args.verbose_filters)
+        stats.report()
+        if manifest_entries:
+            write_manifest(manifest_entries, args.output_dir)
+        return 0 if total else 1
+
+    if args.files:
+        total = 0
+        for path in args.files:
+            if not os.path.isfile(path):
+                log.error(f"File not found: {path}")
+                continue
+            total += extract_from_file(path, prefix, postfixes, args.output_dir, stats, args.verbose_filters)
+        stats.report()
+        if manifest_entries:
+            write_manifest(manifest_entries, args.output_dir)
+        return 0 if total else 1
+
+    parser.print_help()
+    return 2
+
+if __name__ == "__main__":
+    raise SystemExit(main())
\ No newline at end of file