Security audit

Social Media Kit

Security checks across malware telemetry and agentic risk

Overview

The package is labeled as a social media content skill, but the included documentation and executable code are a landing-page generator, so it needs review before installation.

Do not install this as a social media content skill. Only consider it after the publisher corrects the manifest, README, examples, references, and script entrypoint to describe one coherent capability, and review generated HTML/output paths before use in automated workflows.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (7)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 88% confidence
Finding: The skill advertises file read/write behavior through its documented commands and outputs, but no permissions are declared. That creates a trust and sandboxing problem: users or hosting systems may approve the skill under the assumption it is non-privileged, while it can still read input files and write generated artifacts such as JSON and HTML to disk.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 80% confidence
Finding: A description-behavior mismatch is a real security concern because it undermines informed consent and review: the skill claims to generate social media content, but the analyzed behavior indicates broader website/landing-page generation with standalone HTML/CSS/JS output. Hidden or undeclared capabilities increase the risk that users execute a skill that performs materially different actions than expected, especially when it writes active web content to disk.

Description-Behavior Mismatch

High

Confidence: 98% confidence
Finding: The README identifies the skill as a landing page builder, while the manifest claims it is a social media content generator. This kind of capability mismatch is dangerous because reviewers and users may authorize or invoke the skill under false assumptions, increasing the risk of unintended file generation, HTML output handling, or hidden behavior outside the declared purpose.

Description-Behavior Mismatch

High

Confidence: 97% confidence
Finding: The documented feature set focuses on generating production HTML pages, SEO tags, responsive layouts, and web UI components instead of social media assets. In an agent ecosystem, this discrepancy can mislead policy enforcement, tool selection, and user consent, making the skill more dangerous because it may produce or manipulate artifacts outside the expected trust boundary.

Description-Behavior Mismatch

High

Confidence: 99% confidence
Finding: The CLI examples, flags, JSON schema, and output description all describe building standalone HTML landing pages rather than generating social media content. This is a true security-relevant integrity issue because operators may pass data to a skill expecting harmless content planning, while the skill is actually positioned to create web content and files, indicating either a repackaged skill or mislabeled functionality.

Description-Behavior Mismatch

High

Confidence: 99% confidence
Finding: The file implements a full landing-page generator that is materially inconsistent with the declared skill purpose of generating social media content. In an agent skill ecosystem, this kind of capability mismatch is dangerous because it expands the skill's effective behavior beyond what operators expect, enabling unauthorized content generation or deceptive repurposing of the skill.

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The code contains broad HTML/CSS/JS website-generation functionality unrelated to a social-media kit, increasing attack surface and creating an opportunity for misuse under a misleading skill identity. Even though the code does not show overtly malicious payloads, hidden or unjustified capabilities in an agent skill undermine least privilege and trust boundaries.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.