Back to skill

Security audit

Video Analyzer

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because it reports disabled HTTPS certificate checks, a persistent pip package-source change, and other security and reliability concerns; the scan was partial.

Consider avoiding installation until the certificate checks and package-source instructions have been reviewed and the advertised features verified. The scan found real security and reliability concerns but does not establish malicious intent; coverage was partial, with 72 of 74 components scanned and two documentation files only partially inspected.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
core/input_handler.py
Finding

Disabled TLS verification permits HTTPS media substitution

Content
View full analysis

Impact

An attacker can replace the media used for transcription and analysis, corrupt resulting outputs, and persist substituted content in the download cache. This does not establish arbitrary code execution or credential theft.

Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
Findings (95)

Dependency Source Redirection

High
Category
Supply Chain
Confidence
100% confidence
Finding

Dependency resolution is redirected away from a canonical default, adds another source, or uses a destination that cannot be resolved statically.

Content

Scanner excerpt · README.md (reported line 43)May include surrounding context.

md
pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The overall intended purpose matches media analysis, and no unrelated malicious behavior is evident. However, material advertised behavior differs from the implementation. In main(), platform_analysis, platform_meta, editing_result, and several other locals are passed to reporter.generate() before assignment, causing an UnboundLocalError on ordinary runs and preventing later workflows from executing. The --live-analyze block explicitly starts with video_source=None and feeds existing transcript segments with artificial delays, not streaming ASR. In quick mode, the transcript is computed but never written or supplied to TimestampedSummary.generate(), which receives only None and an empty scenes object. Imported module internals were not supplied, so their detailed capabilities cannot be verified.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The setup guidance instructs users to persistently replace pip's default package index, which affects operations beyond this toolkit. This is an unnecessarily broad environment change for a local media-analysis task; the source shows a recommendation, not proof of automatic execution.

Content

No source excerpt is available for this finding.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
85% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · SKILL.md (reported line 307)May include surrounding context.

md
### 运行错误

**Q: 运行中提示 "No space left on device"?**
A: 缓存目录占满了磁盘。用 `--temp-dir` 指定其他目录,或删除旧缓存 `rm -rf .cache/`。也可在 `config.yaml` 中减小 `max_memory_gb`。

**Q: 下载在线视频失败?**
A: 工具内置 3 次重试机制。如果仍然失败,请确认 URL 可访问且网络稳定;YouTube/B站等需安装 yt-dlp:`pip install yt-dlp`。

Direct Prompt Extraction

High
Category
System Prompt Leakage
Confidence
85% confidence
Finding

Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.

Content

Scanner excerpt · core/editing/subtitle_translator.py (reported line 307)May include surrounding context.

python
待翻译文本:
{text_list}"""

        return prompt

    def _call_router(self, router_cmd: List[str], prompt: str) -> Optional[Dict]:
        """调用 cn-llm-router 进行翻译"""

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · core/input_handler.py (reported line 35)May include surrounding context.

python
}
# 系统缓存 / 隐藏文件
FORBIDDEN_SYSTEM_EXTENSIONS = {
    ".ds_store", ".env", ".log", ".tmp",
}
# 其他风险脚本
FORBIDDEN_SCRIPT_EXTENSIONS = {

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
100% confidence
Finding

The --no-check-certificates option disables HTTPS server-certificate validation for yt-dlp. A network attacker could impersonate remote services and substitute downloaded media or metadata, undermining the integrity of the analysis pipeline.

Content

Scanner excerpt · core/input_handler.py (reported line 241)May include surrounding context.

python
"yt-dlp",
                "-f", "best[ext=mp4]/best",
                "-o", output_path,
                "--no-check-certificates",
                "--quiet",
                "--no-warnings",
                url

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
99% confidence
Finding

The --no-check-certificates option disables server certificate validation for yt-dlp network requests. An attacker able to intercept those connections could substitute downloaded media or platform responses, compromising the integrity of files subsequently processed by the local analysis pipeline.

Content

Scanner excerpt · core/platform_adapters/bilibili.py (reported line 114)May include surrounding context.

python
cmd = [
            "yt-dlp",
            "-o", output_path,
            "--no-check-certificates",
            "--quiet",
            "--no-warnings",
            "--merge-output-format", "mp4",

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
99% confidence
Finding

Metadata extraction explicitly disables TLS certificate validation, allowing a network-positioned attacker to impersonate the platform. Forged titles, descriptions, URLs, and statistics could then enter the skill's reports and downstream processing despite the local-analysis focus.

Content

Scanner excerpt · core/platform_adapters/bilibili.py (reported line 154)May include surrounding context.

python
"yt-dlp",
            "--dump-json",
            "--no-download",
            "--no-check-certificates",
            "--quiet",
            video_url,
        ]

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
100% confidence
Finding

Disabling certificate verification allows yt-dlp to accept an unauthenticated HTTPS endpoint during video downloads. A network attacker could substitute media or download responses, despite the URL initially naming Douyin.

Content

Scanner excerpt · core/platform_adapters/douyin.py (reported line 106)May include surrounding context.

python
cmd = [
            "yt-dlp",
            "-o", output_path,
            "--no-check-certificates",
            "--quiet",
            "--no-warnings",
            video_url,

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
100% confidence
Finding

The metadata request disables HTTPS certificate verification even though its output populates trusted analysis fields. A network attacker could alter titles, statistics, attribution, and media URLs, corrupting downstream analysis without needing to modify the original Douyin content.

Content

Scanner excerpt · core/platform_adapters/douyin.py (reported line 141)May include surrounding context.

python
"yt-dlp",
            "--dump-json",
            "--no-download",
            "--no-check-certificates",
            "--quiet",
            video_url,
        ]

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
99% confidence
Finding

Disabling certificate validation removes authentication of the HTTPS endpoints contacted during downloads. A network-positioned attacker could replace media or extraction responses, supplying attacker-controlled content to the local video-processing pipeline.

Content

Scanner excerpt · core/platform_adapters/kuaishou.py (reported line 75)May include surrounding context.

python
cmd = [
            "yt-dlp",
            "-o", output_path,
            "--no-check-certificates",
            "--quiet",
            "--no-warnings",
            video_url,

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
99% confidence
Finding

The metadata request bypasses TLS certificate checks even though it contacts an external platform. An attacker intercepting the connection could inject false metadata and download URLs, undermining analysis and report integrity.

Content

Scanner excerpt · core/platform_adapters/kuaishou.py (reported line 109)May include surrounding context.

python
"yt-dlp",
            "--dump-json",
            "--no-download",
            "--no-check-certificates",
            "--quiet",
            video_url,
        ]

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
99% confidence
Finding

The download command disables TLS certificate verification, permitting server impersonation by an attacker with network interception capability. Substituted media or extraction responses would be accepted into a workflow that subsequently processes downloaded files locally.

Content

Scanner excerpt · core/platform_adapters/wechat_video.py (reported line 88)May include surrounding context.

python
cmd = [
            "yt-dlp",
            "-o", output_path,
            "--no-check-certificates",
            "--quiet",
            "--no-warnings",
            video_url,

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
99% confidence
Finding

Certificate verification is disabled for metadata retrieval, so HTTPS no longer reliably authenticates the remote platform. A network-positioned attacker could alter metadata and URLs that the skill stores and uses in subsequent analysis.

Content

Scanner excerpt · core/platform_adapters/wechat_video.py (reported line 122)May include surrounding context.

python
"yt-dlp",
            "--dump-json",
            "--no-download",
            "--no-check-certificates",
            "--quiet",
            video_url,
        ]

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · README.md (reported line 16)May include surrounding context.

md
# 2. 安装 ffmpeg(必须)
# Windows: scoop install ffmpeg  أو  choco install ffmpeg
# macOS: brew install ffmpeg
# Ubuntu/Debian: sudo apt install ffmpeg

# 3. 验证安装
ffmpeg -version

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 278)May include surrounding context.

md
### 安装与配置

**Q: ffmpeg 安装后仍提示找不到?**
A: 需要让 ffmpeg 在系统 PATH 中。Windows 用户可用 `scoop install ffmpeg` 或下载后解压并将 bin 目录加入 PATH;macOS 用 `brew install ffmpeg`;Ubuntu 用 `sudo apt install ffmpeg`。安装后新开一个终端窗口,运行 `ffmpeg -version` 验证。

**Q: pip install 时 whisper 下载模型很慢?**
A: whisper 首次运行时会自动下载模型文件(small 约 466MB),国内用户建议设置镜像:`pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple`

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · core/doctor.py (reported line 326)May include surrounding context.

python
elif platform.system() == "Darwin":
            return "macOS 安装 ffmpeg: brew install ffmpeg"
        else:
            return "Linux 安装 ffmpeg: sudo apt install ffmpeg 或 sudo dnf install ffmpeg"

    def _generate_summary(self, report: Dict, fixes_needed: List[str]) -> str:
        """生成自检摘要"""

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · core/doctor.py (reported line 326)May include surrounding context.

python
elif platform.system() == "Darwin":
            return "macOS 安装 ffmpeg: brew install ffmpeg"
        else:
            return "Linux 安装 ffmpeg: sudo apt install ffmpeg 或 sudo dnf install ffmpeg"

    def _generate_summary(self, report: Dict, fixes_needed: List[str]) -> str:
        """生成自检摘要"""

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · core/editing/bgm_recognizer.py (reported line 300)May include surrounding context.

python
"安装方法:\n"
            "  Windows: choco install chromaprint\n"
            "  macOS: brew install chromaprint\n"
            "  Linux: sudo apt install chromaprint-fpcalc\n"
            "安装后重试即可启用 BGM 识别功能。"
        )

The download workaround changes pip's global package index without warning about its persistent, broader impact.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The workaround persistently changes pip's default package index, potentially affecting unrelated environments and future installations that read this configuration. It expands package-source trust without explaining the scope, although there is no evidence that the named mirror is malicious.

Content

No source excerpt is available for this finding.

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.

The setup instructions recommend a persistent pip package-source change without disclosing its effect on unrelated installations.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The first-run guidance presents a persistent pip index replacement as a download acceleration step without disclosing its effect on unrelated package installations. Changing pip's index also does not generally redirect Whisper model-file downloads, making this broader trust change unnecessary for the stated problem.

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 · core/chapter_slicer.py (reported line 143)May include surrounding context.

python
]
        
        try:
            result = subprocess.run(cmd, capture_output=True, timeout=120)
            if result.returncode != 0:
                # 如果复制流失败(可能是关键帧问题),回退到重新编码
                return self._extract_video_clip_reencode(video_path, start, duration, output_path)

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 · core/chapter_slicer.py (reported line 179)May include surrounding context.

python
]
        
        try:
            result = subprocess.run(cmd, capture_output=True, timeout=300)
            return result.returncode == 0 and os.path.exists(output_path)
        except Exception as e:
            logger.debug(f"重新编码切割失败: {e}")