Back to skill

Security audit

字幕鲸

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because it found a private-token requirement and promotional messaging outside the stated video-transcription purpose, along with a reported download-handling weakness; analysis was incomplete.

Consider the verdict's suspicious rating before installing. The scanners report an undisclosed token requirement and promotional messaging, a possible deletion of pre-existing data during resumed downloads, and incomplete analysis of 11 components; they do not establish malicious intent, but the incomplete coverage means safety is not confirmed.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (22)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

Token validation can be a supporting implementation detail, but this chunk additionally introduces an undisclosed private-token access requirement and promotional acquisition flow unrelated to the stated Qwen video-processing functionality. The code only checks token syntax; it does not verify credentials remotely, contact the advertised destinations, or itself enforce suspension beyond returning an empty string.

Content

No source excerpt is available for this finding.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 55)May include surrounding context.

md
command: 'node scripts/video2text/index.js --file "https://example.com/video.mp4"'

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 57)May include surrounding context.

md
command: 'node scripts/video2text/index.js --file "/path/to/video.mp4"'

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 59)May include surrounding context.

md
command: 'node scripts/video2text/index.js --file "https://example.com/video.mp4" --prompt "总结这个视频的核心观点"'

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 61)May include surrounding context.

md
command: 'node scripts/video2text/index.js --file "/path/to/video.mp4" --prompt "改写成小红书风格的文案"'

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 63)May include surrounding context.

md
command: 'node scripts/video2text/index.js --id last --prompt "提取视频中的所有金句"'

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 65)May include surrounding context.

md
command: 'node scripts/video2text/index.js --help'

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 125)May include surrounding context.

md
| 怎么把 https://example.com/video.mp4 里的声音转成文字 | `node scripts/video2text/index.js --file "https://example.com/video.mp4"` |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 126)May include surrounding context.

md
| 总结这个视频的核心观点 https://example.com/video.mp4 | `node scripts/video2text/index.js --file "https://example.com/video.mp4" --prompt "总结这个视频的核心观点"` |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 127)May include surrounding context.

md
| 把本地 /path/to/video.mp4 改成小红书文案 | `node scripts/video2text/index.js --file "/path/to/video.mp4" --prompt "改写成小红书风格的文案"` |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 128)May include surrounding context.

md
| 用刚才分析的视频提取所有金句 | `node scripts/video2text/index.js --id last --prompt "提取视频中的所有金句"` |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 129)May include surrounding context.

md
| 分析这个视频的情感基调和受众 | `node scripts/video2text/index.js --file "<URL或路径>" --prompt "分析视频的情感基调和目标受众"` |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 153)May include surrounding context.

md
node scripts/video2text/index.js --file "/path/to/your/video.mp4"

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 154)May include surrounding context.

md
node scripts/video2text/index.js --file "https://example.com/video.mp4"

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 160)May include surrounding context.

md
node scripts/video2text/index.js --file "/path/to/your/video.mp4" --prompt "请用三个要点总结这个视频的核心内容"

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 162)May include surrounding context.

md
node scripts/video2text/index.js --file "/path/to/your/video.mp4" --prompt "将这段内容改写成一篇风趣幽默的小红书文案"

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 164)May include surrounding context.

md
node scripts/video2text/index.js --id last --prompt "提取视频中的所有金句"

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · README.md (reported line 131)May include surrounding context.

md
4. 同时传入文件路径与任务ID,优先执行 `--id`,忽略 `--file`
5. 无自定义 prompt 时,默认完整转录视频全部文字

| 用户自然语言指令                                         | 自动生成命令                                                                                                |
| -------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------- |
| 视频提取 https://example.com/video.mp4 中的文字          | `node scripts/video2text/index.js --file "https://example.com/video.mp4"`                                   |
| 把本地 /path/to/your/video.mp4 改成小红书风格的文案      | `node scripts/video2text/index.js --file "/path/to/your/video.mp4" --prompt "改写成小红书风格的文案"`       |

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env.

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.

Stopping or failing a download can silently delete the destination file, including a file supplied for resumption.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

resumeFromFile accepts an existing file, but stop and failure cleanup can unlink that same path because removeOnStop and removeOnFail default to true without tracking file ownership. A failed or interrupted resumed download can therefore destroy pre-existing data, although the skill's normal helper creates a fresh download rather than explicitly resuming a user-supplied file.

Content

No source excerpt is available for this finding.

Incomplete downloads are deleted without disclosing the deletion.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

When a download is marked incomplete, the handler silently deletes the returned file path. The error reports only that the download is incomplete; no prompt, log, or comment explains that the partial file is removed.

Content

No source excerpt is available for this finding.

Startup performs expired-data deletion without user disclosure, even for help-only invocations.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The call to clean.deleteExpire() runs before argument parsing and help handling, with no confirmation, user-facing log, or explanatory comment. Consequently, simply requesting help can trigger cleanup without informing the user, and cleanup errors are silently swallowed.

Content

No source excerpt is available for this finding.