Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed local video-analysis helper that measures a finished video and exports a reusable production reference without evidence of hidden network, persistence, credential, or destructive behavior.
Install this only for a workflow where a local agent may run Python plus ffmpeg/ffprobe on videos you choose. Keep the work directory local because intermediate breakdowns and frame sheets can contain source-video facts and imagery; share only the exported production_reference.json after checking it.
The implementation is a frame-sheet utility for visual inspection of video timing/cuts. That may support analysis, but it does not implement the declared tool behavior of creating a reusable production reference JSON with higher-level annotations and separated facts/methods. The mismatch is in primary purpose and delivered artifact, not just missing minor details.
The skill explicitly instructs the agent to run local Python and ffmpeg-based commands that read and write files and invoke a shell, but it does not declare any permissions or allowed-tools scope. This creates an overbroad execution surface where a host agent may grant more capability than intended, increasing the chance of unauthorized file access or command execution if the skill or surrounding workflow is abused.
SQP-3 applies to all file types and covers language or locale policy violations. This file begins with a Chinese-only title and the rest of the instructions are also Chinese, with no opt-in, alternative language, or justification for a locale-specific audience.
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
def run_cmd(cmd, **kwargs):
"""Run a command list and return the CompletedProcess (stdout/stderr captured)."""
return subprocess.run(
[str(part) for part in cmd], capture_output=True, text=True,
encoding="utf-8", errors="replace", **kwargs,
)
The code raises RuntimeError messages in Chinese ('失败', '没有视频流', '无法读取时长'), which imposes a specific language on users. Under the policy, forcing a language without opt-in or documented justification is a natural-language policy violation.
This code includes user-facing labels in Chinese, such as the row label at L034, and similar Chinese-only messages elsewhere in the file. The file provides no indication that the skill is intentionally region-specific or that users can choose their preferred language, which violates the language/locale policy for natural-language content.
The ValueError message at L050 is written only in Chinese, which forces a specific language for error handling. There is no surrounding mechanism for locale selection or documentation justifying a Chinese-only audience.
The RuntimeError text at L067 is user-visible and written only in Chinese. Because the file does not offer language selection or explain a locale-specific constraint, this is a natural-language policy violation.
The ffmpeg page-render failure message at L093 is Chinese-only and user-visible. Without any opt-in, localization support, or documented regional scope, this hard-codes a language choice in violation of policy.
The legend line generated at L104 contains Chinese text and is part of normal output presented to users. The skill does not provide a language choice or state that it is restricted to a Chinese-language environment.
The ValueError at L115 is user-facing and written only in Chinese. There is no evidence of localization support or a documented Chinese-only scope, so this is a language policy issue.
This code emits user-visible error and log messages only in Chinese, including the ValueError text and status logs. The file does not indicate that the skill is region-specific or that users can choose their language, which creates a locale policy issue under the natural-language policy rules.
This Python file contains user-facing status messages and command descriptions in Chinese only, such as the printed report text and the ArgumentParser description. The policy forbids forcing a specific language without user opt-in, and the file provides no option to select locale or justify a region-specific restriction.
No suspicious patterns detected.