Back to skill

Security audit

UAvideo

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this video analysis skill suspicious because it found a confirmed risk that temporary downloads may overwrite existing files; the overall scan confidence is medium and some coverage was incomplete.

Consider holding off on installing until the temporary file overwrite issue is reviewed or fixed. The malware signature and subprocess flags have conflicting or unconfirmed assessments, so they do not establish malicious intent; however, the scan was partial: a local reference was missing and binary content was outside some analyzer coverage.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (11)

YARA rule 'info_stealer': Information stealer patterns (credential harvesting, browser data theft) [malware]

High
Category
YARA Match
Confidence
75% confidence
Finding

YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).

Content

Scanner excerpt · ENVIRONMENT-NOTES.md (reported line 54)May include surrounding context.

s\download_media.py "https://v.douyin.com/xxxx/" -o out

text

部分平台的 extractor 要求请求携带新鲜 Cookie(`ttwid` / `msToken` / `odin_tt` 等字段组合,无需登录态)。仅提供单一 Cookie 字段可能仍被拦截,表现为 `Fresh cookies are needed` 类错误。

### 路径 2:`--cookies-from-browser`(受限环境通常不可行)

```powershell
python -m yt_dlp --cookies-from-browser chrome ...

该路径存在两类独立障碍:浏览器运行期间 Cookie 数据库被独占锁定;新版本浏览器的 app-bound encryption(Cookie 加密绑定浏览器进程)使 Cookie 难以被外部工具读取。除非用户明确愿意关闭全部浏览器进程并接受仍可能失败,否则不建议在此路径上投入排查时间。

路径 3:平台移动端接口解析(推荐的降级方案)

原理:通过短链跳转获取视频 aweme_id,再调用平台移动端 feed 接口取得详情 JSON,从中提取 video.play_addr / `bit

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: file_read, file_write, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Temporary downloads can overwrite existing files despite the documented no-overwrite guarantee.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

unique_path checks only the final destination, while curl and FFmpeg write to predictable temporary paths that may already exist; FFmpeg explicitly enables overwriting with '-y'. This can destroy existing temporary files or corrupt concurrent downloads, and a pre-positioned symlink can redirect writes to another writable file where the platform permits it.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/download_media.py (reported line 47)May include surrounding context.

python
dest = unique_path(out_dir / name)
    tmp = dest.with_suffix(dest.suffix + ".part")
    cmd = ["curl", "-L", "--fail", "--retry", "3", "-o", str(tmp), url]
    subprocess.run(cmd, check=True)
    shutil.move(str(tmp), str(dest))
    return dest

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/download_media.py (reported line 56)May include surrounding context.

python
dest = unique_path(out_dir / "stream.mp4")
    tmp = dest.with_suffix(".part.mp4")
    cmd = ["ffmpeg", "-y", "-i", url, "-c", "copy", "-bsf:a", "aac_adtstoasc", str(tmp)]
    subprocess.run(cmd, check=True)
    shutil.move(str(tmp), str(dest))
    return dest

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/download_media.py (reported line 69)May include surrounding context.

python
sys.exit(2)
    dest_tpl = str(out_dir / "douyin_%(id)s_%(title).40s.%(ext)s")
    cmd = ["yt-dlp", "--no-playlist", "-o", dest_tpl, url]
    subprocess.run(cmd, check=True)
    # 找最新的文件
    files = sorted(out_dir.glob("*"), key=lambda p: p.stat().st_mtime, reverse=True)
    return files[0] if files else out_dir

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/extract_frames.py (reported line 43)May include surrounding context.

python
ffprobe = find_bin("ffprobe")
    cmd = [ffprobe, "-v", "quiet", "-print_format", "json",
           "-show_format", "-show_streams", str(video)]
    info = json.loads(subprocess.run(cmd, capture_output=True, text=True, check=True).stdout)
    fmt = info.get("format", {})
    vstream = next((s for s in info.get("streams", []) if s.get("codec_type") == "video"), {})
    dur = float(fmt.get("duration", 0) or 0)

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/extract_frames.py (reported line 63)May include surrounding context.

python
ffmpeg = find_bin("ffmpeg")
    cmd = [ffmpeg, "-y", "-i", str(video), "-vf", f"fps=1/{interval},scale=640:-2",
           "-q:v", "3", str(out_dir / "f_%02d.jpg")]
    subprocess.run(cmd, check=True, capture_output=True)
    n = len(list(out_dir.glob("f_*.jpg")))
    return n

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/extract_frames.py (reported line 99)May include surrounding context.

python
seg_path = seg_dir / name
        cmd = ["ffmpeg", "-y", "-ss", str(s), "-to", str(e), "-i", str(video),
               "-c", "copy", str(seg_path)]
        subprocess.run(cmd, check=True, capture_output=True)
        # 3) 段抽帧: 首帧+中点帧
        for tag, at in (("a", s + 0.5), ("b", (s + e) / 2)):
            cmd = ["ffmpeg", "-y", "-ss", str(at), "-i", str(video),

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/extract_frames.py (reported line 78)May include surrounding context.

python
# 1) 场景切点检测
    detect = [ffmpeg, "-i", str(video), "-vf",
              "select='gt(scene,0.4)',showinfo", "-f", "null", "-"]
    r = subprocess.run(detect, capture_output=True, text=True)
    cuts = []
    for line in r.stderr.splitlines():
        if "pts_time:" in line and "showinfo" in line:

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/extract_frames.py (reported line 105)May include surrounding context.

python
cmd = ["ffmpeg", "-y", "-ss", str(at), "-i", str(video),
                   "-frames:v", "1", "-q:v", "3",
                   str(frame_dir / f"{seg_path.stem}_{tag}.jpg")]
            subprocess.run(cmd, check=True, capture_output=True)
        segments.append({"name": name, "start": s, "end": e, "dur": round(e - s, 1)})
    return segments