ClawHub
Back to skill

Security audit

Markdown Linter

Security checks across malware telemetry and agentic risk

Overview

This appears to be a Markdown linting skill with an optional external-link checker; the network behavior is relevant to the purpose but users should enable it deliberately.

Install is reasonable if you want Markdown linting. Keep external link checking off unless you intentionally want network validation, especially when linting third-party or private Markdown files, because embedded links may be contacted from your environment.

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 (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding
The skill declares no permissions, but its documented behavior and metadata indicate file reading and optional network access for external link validation. This creates a capability/permission mismatch that can mislead operators and policy engines, reducing visibility into what the skill can access and making unintended data exposure or outbound requests more likely.

Description-Behavior Mismatch

Medium
Confidence
90% confidence
Finding
The code optionally performs outbound HTTP requests to every external URL found in markdown content. In an agent skill advertised as a Markdown linter, this expands the trust boundary from local file parsing to network interaction, which can leak network metadata, contact attacker-controlled hosts, and create SSRF-style behavior if untrusted markdown is linted inside a sensitive environment.

Context-Inappropriate Capability

Medium
Confidence
85% confidence
Finding
The constructor and runtime flow expose an optional network-capable feature that is not clearly necessary for a local formatting/style linter. Even though it is opt-in, this can still be dangerous in agent environments because users or orchestrators may enable it without understanding that linting untrusted markdown will trigger outbound requests to attacker-supplied URLs.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.