Security audit

You Get

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent media-downloader guide, but it gives an AI broad command authority and includes sensitive cookie, proxy, sudo, and firewall-disabling workflows without enough safety controls.

Install only if you are comfortable with an AI helper running download and package-management commands. Do not paste raw cookies into chat, do not allow automatic access to browser cookie databases, review every URL and output path before execution, avoid untrusted proxies, and reject any firewall-disabling or sudo action unless you explicitly intend it.

SkillSpector

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
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (13)

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: The guide explicitly authorizes AI to launch local media players and browsers, expanding the skill from media downloading into execution of external applications. This increases the attack surface because an agent may open untrusted URLs or invoke local applications with attacker-influenced arguments, which is riskier than passive downloading alone.

Context-Inappropriate Capability

Medium

Confidence: 92% confidence
Finding: The batch-download section tells the AI it may create shell scripts and run shell automation, which broadens the capability from a downloader into general-purpose file creation and command orchestration. In an agent setting, this can be abused to write files, persist instructions, or execute attacker-shaped shell flows beyond the intended scope.

Context-Inappropriate Capability

Low

Confidence: 79% confidence
Finding: The Google search-based retrieval feature adds an open-ended search capability beyond downloading a user-specified media URL. In an agent context, this can lead to unintended navigation, retrieval from untrusted sources, and scope creep from downloader to web-search actor.

Description-Behavior Mismatch

Medium

Confidence: 94% confidence
Finding: The documentation extends use from media downloads to arbitrary file downloads, which materially broadens the skill beyond its stated purpose. In an agent environment, arbitrary file retrieval can be leveraged to fetch malware, scripts, or sensitive content from untrusted locations.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill explicitly advertises 'cookies登录下载' without any guidance on how authentication cookies should be sourced, stored, redacted, or avoided in shared/agent logs. In an AI-assisted workflow that may ask users to paste cookies or run commands containing credential material, this can lead to account/session theft, unauthorized access, or leakage of sensitive authentication data through prompts, shell history, or tool output.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The guide explicitly tells the AI it can directly execute package installation commands (`pip install you-get`) and verification commands without a warning or confirmation step. In an agent setting, this can cause unintended system modification, dependency changes, or installation of unreviewed code from external package registries, which is risky even if the underlying package is legitimate.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The document includes privileged package-management commands such as `sudo apt install ffmpeg` and `sudo yum install ffmpeg` without highlighting elevated-permission risks. In an autonomous or semi-autonomous agent workflow, this could trigger system-wide changes with administrator privileges, increasing the blast radius if executed on the wrong host or without user awareness.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The guide explicitly tells an AI to directly execute a media download command against an external URL. That causes network access and local file creation without any warning, approval gate, or confirmation step, which is unsafe for an autonomous agent context even if the command itself is not inherently malicious.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The save-path section instructs the AI that it may customize output locations and filenames, including writing into user directories such as ~/Videos. In an agent setting, this can result in unintended file writes, overwrites, or clutter in sensitive locations if the path is inferred or user-controlled without validation and consent.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The guide instructs use of cookies.txt and browser cookie databases without warning that these files contain authenticated session material. In an AI-assisted workflow, this can expose sensitive credentials or enable account misuse if the agent is allowed to read, store, or transmit those artifacts.

Missing User Warnings

Medium

Confidence: 85% confidence
Finding: The proxy guidance encourages routing traffic through third-party infrastructure, including for blocked or region-restricted content, without disclosing privacy and integrity risks. This can expose browsing targets, downloads, and metadata to untrusted intermediaries, especially when an agent is configured to use arbitrary proxies.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The troubleshooting guidance tells users to temporarily disable the firewall to test connectivity, but it does not include a clear warning about increased system exposure, scope limitation, or immediate re-enablement steps. In a command-oriented support document, users may copy the command directly, which can weaken host defenses and create avoidable exposure during troubleshooting.

Ssd 3

Medium

Confidence: 88% confidence
Finding: The document promotes using browser cookies to access membership-only or age-restricted content, normalizing the use of authenticated session artifacts to bypass content gates. In an agent context, this raises abuse and account-misuse risk because the agent may handle privileged cookies for restricted resources.

Static analysis

No suspicious patterns detected.