Back to skill

Security audit

Video Subtitle Translation & Dubbing

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches its video translation and dubbing purpose, but it should be reviewed because it can send subtitle text, generated speech text, and bearer credentials to user-supplied endpoints without secure destination checks.

Review the configured translation and TTS providers before installing or running this skill. Use HTTPS endpoints you control or trust, avoid processing confidential videos unless the provider's retention and privacy terms are acceptable, and use limited-scope API keys with separate credentials for translation and TTS where possible.

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

Error
Location
src/translation_dubbing_skill/entry/manifest.py:441
Finding

Bearer Credentials and Subtitle Content Can Be Sent to Arbitrary or Plaintext Endpoints

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (74)

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 · README.md (reported line 66)May include surrounding context.

3. 配置凭证

复制环境变量模板文件并填写你的 API Keys:

bash
cp .env.example .env

根据 .env 中的说明,配置你的大模型或翻译/TTS 提供商凭证。

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 · README.zh-CN.md (reported line 66)May include surrounding context.

3. 配置凭证

复制环境变量模板文件并填写你的 API Keys:

bash
cp .env.example .env

根据 .env 中的说明,配置你的大模型或翻译/TTS 提供商凭证。

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description promises a multi-language subtitle translation and automatic dubbing skill. The supplied code does not translate text, does not call any language model or translation service, does not generate audio, and does not perform dubbing. Its actual function is narrowly focused on preprocessing subtitle files: parsing VTT/SRT, removing inline timestamps/tags, deduplicating rolling captions, segmenting text into sentence-level cues, and serializing cleaned subtitles back to VTT/SRT. While such cleaning could be a supporting step in a translation/dubbing pipeline, this code chunk itself does not implement the declared core capabilities, so the description materially misrepresents the behavior.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The supplied code clearly matches part of the description: subtitle translation, including configurable source/target languages, batching, and robust parsing of LLM responses. However, this chunk does not implement any dubbing behavior, audio synthesis, voice generation, or video processing. Its sole responsibility is text translation of subtitle entries. That makes the declared description materially broader than the actual behavior shown here. Additionally, although this may be normal implementation detail for an LLM provider, the code does use network access and bearer-token authentication despite no declared permissions being listed.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The code clearly implements only a translation provider for subtitle text using an external HTTP translation API. It accepts subtitle entries, translates their text, and returns updated subtitle entries. This aligns partially with the 'subtitle translation' portion of the description, but it does not implement automatic dubbing or any audio/video processing. Because the declared purpose prominently includes automatic dubbing and multi-language video subtitle translation as a skill-level capability, while the supplied code chunk only provides text translation infrastructure, this is a description-behavior mismatch for the chunk evaluated.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared description promises a multi-language video subtitle translation and automatic dubbing skill. The supplied code chunk, however, is narrowly focused on one provider component: an HTTP client for LLM-backed text-to-speech synthesis. It accepts text, posts it to a remote endpoint with authorization, parses audio responses, and maps errors. There is no evidence here of subtitle extraction, translation between languages, video handling, subtitle file manipulation, or orchestration of a complete dubbing workflow. While TTS is consistent with the dubbing portion, the actual code is only a partial implementation and omits major declared capabilities. Additionally, the code clearly performs external HTTP requests with bearer-token authentication, which is a resource access capability not reflected in the declared permissions.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied code chunk does not implement subtitle translation, dubbing, language handling, video processing, or media workflows. Instead, it provides a dataclass and helper validation for scheduler/runtime rate-limit configuration. While such scheduling logic could be a supporting internal component of a larger translation/dubbing system, this specific chunk's behavior is materially different from the declared end-user purpose and exposes undeclared scheduler capabilities as its primary function.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The declared description promises a multi-language subtitle translation and automatic dubbing skill. However, the supplied code chunk implements only subtitle extraction from a video container, specifically trying English subtitle streams via ffmpeg mappings (eng, en, then first subtitle stream) and returning SRT text. It does not translate subtitles, generate dubbed audio, handle multiple target languages, or perform broader dubbing workflow actions. This is a material description-behavior mismatch because the actual code’s primary purpose is limited subtitle extraction rather than translation/dubbing.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description presents a high-level media-processing skill for translating subtitles across multiple languages and automatically dubbing videos. However, the supplied code chunk is narrowly focused on parsing subtitle text formats (SRT/VTT) into structured entries and validating timestamps. This is a supporting component for subtitle workflows, but by itself it does not implement the stated primary capabilities of translation or dubbing. There are no undeclared sensitive permissions or resource accesses, but the actual behavior is materially narrower and different from the declared purpose, so this is a description-behavior mismatch.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description claims a skill for multi-language subtitle translation and automatic dubbing. The supplied code chunk does not perform translation, language detection, audio synthesis, dubbing, or video processing. Its sole purpose is subtitle serialization: converting SubtitleEntry objects into canonical SRT/VTT text and optionally writing that text to a file. While subtitle formatting could be a supporting component within a broader translation/dubbing system, this code chunk itself materially differs from the declared primary purpose and exposes only subtitle pretty-printing behavior.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The supplied code clearly matches part of the description: it translates subtitle entries using providers and scheduling logic. However, the declared purpose says the skill provides both subtitle translation and automatic dubbing for many languages. In this code chunk, there is no dubbing-related functionality such as text-to-speech, voice synthesis, audio track generation, lip-sync, or video/audio manipulation. The implementation is limited to translation coordination and validation. Additionally, while the API accepts arbitrary target_language values, the semantic contract specifically enforces CJK content when target_language starts with 'zh', and the defaults are English-to-Chinese, so the chunk does not substantiate the broad multi-language dubbing claim. Therefore the description overstates the functionality represented by this code chunk.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared description presents a combined subtitle translation and automatic dubbing skill for videos. The supplied code, however, is narrowly focused on TTS orchestration: synthesizing audio clips from subtitle text via a provider and scheduler. It handles batching, rate-limit-aware scheduling, provider initialization, text normalization, progress reporting, and output validation. There is no evidence in this chunk of translation between languages, subtitle translation logic, language detection, or video manipulation/container handling. This is therefore a material description-behavior mismatch for this code chunk: the actual implemented purpose is only the dubbing/TTS engine portion, not the broader multi-language subtitle translation and video dubbing capability as declared.

Content

No source excerpt is available for this finding.

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 · scripts/run_quick_translate.py (reported line 29)May include surrounding context.

python
def load_env() -> None:
    """Simple parser to load variables from a local .env file."""
    env_path = PROJECT_ROOT / ".env"
    if not env_path.exists():
        print("💡 Hint: No .env file found in project root. Reading from system environments.")

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 · scripts/run_quick_translate.py (reported line 32)May include surrounding context.

python
def load_env() -> None:
    """Simple parser to load variables from a local .env file."""
    env_path = PROJECT_ROOT / ".env"
    if not env_path.exists():
        print("💡 Hint: No .env file found in project root. Reading from system environments.")

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 · scripts/run_quick_translate.py (reported line 94)May include surrounding context.

python
def load_env() -> None:
    """Simple parser to load variables from a local .env file."""
    env_path = PROJECT_ROOT / ".env"
    if not env_path.exists():
        print("💡 Hint: No .env file found in project root. Reading from system environments.")

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 · scripts/run_quick_translate.py (reported line 30)May include surrounding context.

python
def load_env() -> None:
    """Simple parser to load variables from a local .env file."""
    env_path = PROJECT_ROOT / ".env"
    if not env_path.exists():
        print("💡 Hint: No .env file found in project root. Reading from system environments.")
        return

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 · scripts/run_quick_translate.py (reported line 86)May include surrounding context.

python
def load_env() -> None:
    """Simple parser to load variables from a local .env file."""
    env_path = PROJECT_ROOT / ".env"
    if not env_path.exists():
        print("💡 Hint: No .env file found in project root. Reading from system environments.")
        return

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The README explicitly instructs users to configure external translation and TTS HTTP endpoints and to provide API credentials, but it does not warn that subtitle text, audio, and potentially sensitive video-derived content may be transmitted to third-party services. In a media-processing skill, subtitles and spoken content can contain confidential or regulated information, so omission of a privacy/data-sharing notice creates a real security and compliance risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README explicitly instructs users to configure remote translation and TTS providers, which strongly implies subtitle text, transcript content, and possibly derived audio will be sent to third-party services. The missing disclosure is a real security/privacy issue because users may process sensitive video content without understanding that data leaves the local environment.

Content

No source excerpt is available for this finding.

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 53)May include surrounding context.

md
- **Python 3.11+**
- **FFmpeg**:需要在系统的 `PATH` 环境变量中可用。
  - **macOS**: `brew install ffmpeg`
  - **Ubuntu**: `sudo apt install ffmpeg`

### 2. 安装依赖
克隆项目后,在根目录下执行:

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.zh-CN.md (reported line 53)May include surrounding context.

md
- **Python 3.11+**
- **FFmpeg**:需要在系统的 `PATH` 环境变量中可用。
  - **macOS**: `brew install ffmpeg`
  - **Ubuntu**: `sudo apt install ffmpeg`

### 2. 安装依赖
克隆项目后,在根目录下执行:

Undeclared Tool Scope

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.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring states that the pipeline runs in 'subtitle_and_dubbing' mode with 'Chinese dub + Chinese subtitles'. This imposes a specific output language/locale as the default behavior, and the script does not expose any user-facing option to choose another language or explicitly opt into Chinese output.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The manifest describes a multi-language video subtitle translation and dubbing skill, which reasonably implies media parsing, subtitle processing, networked translation, and audio generation. However, this helper script additionally invokes an external binary via subprocess to inspect files, introducing process-execution capability that is not stated or obviously required by the declared purpose.

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/batch_run.py (reported line 133)May include surrounding context.

python
import subprocess

    try:
        result = subprocess.run(
            [
                "ffprobe",
                "-v",