Back to skill

Security audit

Video Content Analyzer

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent video-analysis helper, but users should be careful with its broad shell access, external transcription API use, and unsafe dependency install example.

Install dependencies in a virtual environment or with pipx instead of using `--break-system-packages`, review yt-dlp/ffmpeg commands before running them, and only submit videos whose audio/transcript you are comfortable sending to OpenAI Whisper for transcription.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:62
Finding
Unpinned Dependency Installation Bypasses System Package Protections## Vulnerability Details **File Location**: `SKILL.md:62` **Vulnerability Type**: Insecure third-party dependency installation **Risk Level**: Medium **Complete Code Snippet**: ```bash # Install yt-dlp (video download) pip3 install --break-system-packages yt-dlp ``` ### Technical Analysis The documented setup command installs the latest available `yt-dlp` release without pinning a reviewed version or verifying an integrity hash. Package resolution is therefore mutable: running the same command at different times may install different code from the configured Python package index. The `--break-system-packages` option also disables the protection applied to externally managed Python environments. This allows `pip` to alter packages associated with the host's system-managed Python installation, increasing the potential for dependency conflicts and damage to applications that share that environment. Exploitation depends on compromise or manipulation of the package supply chain, such as a compromised upstream release, package-index account, index configuration, or network/package mirror. There is no evidence in the audited file that `yt-dlp` itself is malicious; the finding concerns the unsafe installation procedure. ### Attack Path 1. An attacker compromises a future `yt-dlp` release or a package source used by the environment. 2. An operator follows the documented command. 3. `pip` resolves the mutable latest release without checking a pinned version or expected hash. 4. Package-controlled installation or runtime code executes with the permissions of the user running `pip`. 5. Because system-package protection was explicitly bypassed, the installation may also modify or conflict with the system-managed Python environment. ### Impact Assessment A successful supply-chain compromise could execute arbitrary code with the installing user's privileges. It could access data and credentials available to that user, alter use ...[truncated 278 chars]
Remediation
## Remediation Suggestions - Remove `--break-system-packages`. - Install the dependency in an isolated virtual environment or through `pipx`. - Pin `yt-dlp` to a specifically reviewed version rather than installing the mutable latest release. - Use hash verification through a locked requirements file, for example `pip install --require-hashes -r requirements.txt`. - Review and deliberately update the pinned version on a controlled schedule. - Run dependency installation and video processing as an unprivileged user in a constrained environment. - Prefer a trusted internal package mirror with provenance, integrity, and release-approval controls where available. A hardened installation pattern is: ```bash python3 -m venv .venv . .venv/bin/activate python3 -m pip install --require-hashes -r requirements.txt ``` The requirements file should contain an approved exact version and its expected distribution hashes.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Natural-Language Policy Violations

Medium
Confidence
92% confidence
Finding
The manifest restricts languages to `zh`, which imposes a specific language/locale constraint. There is no indication that users can opt into this limitation or that the constraint is justified as a region-specific tool, so it appears to violate the language/locale policy.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Content
# 安装 ffmpeg (音频处理)
# Mac: brew install ffmpeg
# Ubuntu: sudo apt install ffmpeg
```

### API Key
Confidence
70% confidence
Finding
Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill sends user-provided audio/video-derived content to the external OpenAI Whisper API, but the user-facing description and workflow do not clearly warn that third-party processing occurs or that potentially sensitive spoken content may leave the local environment. This creates a real privacy and consent risk, especially for proprietary, personal, or regulated media.

Static analysis

No suspicious patterns detected.