Security audit

cdn-toolkit-free

Security checks across malware telemetry and agentic risk

Overview

The skill is CDN-focused, but it includes live CDN API changes and host firewall commands without enough scoping, warnings, or rollback guidance.

Install only if you want an agent to help with CDN diagnostics and configuration, and treat all Cloudflare API and firewall snippets as manual-review examples. Do not let an agent run the PATCH or iptables commands against production without a scoped token, maintenance plan, backup/rollback path, and explicit confirmation.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands

Findings (7)

Description-Behavior Mismatch

High

Confidence: 93% confidence
Finding: The skill states it is not suitable for operations or deployment management, yet elsewhere includes executable commands that change live infrastructure and host configuration. This mismatch can cause the agent or user to invoke the skill in broader contexts than intended, leading to unintended system changes under a misleading safety boundary.

Context-Inappropriate Capability

High

Confidence: 97% confidence
Finding: The skill includes iptables commands that directly alter host firewall policy by permitting Cloudflare source ranges and then dropping other HTTPS traffic. In an agent context with exec access, this can lock out legitimate traffic, break origin access, or cause self-inflicted denial of service if applied incorrectly or on the wrong host.

Intent-Code Divergence

Medium

Confidence: 90% confidence
Finding: The documentation frames the free edition as providing configuration guidance, but included examples perform real remote state changes through CDN provider APIs. That discrepancy is dangerous because users may treat the skill as informational while it actually encourages authenticated modification of production CDN settings.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The trigger conditions are overly broad, covering generic coding, debugging, and deployment requests, which increases the chance the skill will activate outside its narrow CDN-focused use case. In a skill with exec capability and infrastructure-changing examples, over-triggering materially raises the risk of irrelevant or unsafe command suggestions being surfaced in unrelated workflows.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The Cloudflare examples use API tokens to send authenticated PATCH requests that modify live CDN settings, yet the document lacks strong warnings about production impact, token handling, rollback, and least-privilege scoping. In an agent-assisted environment, such examples can normalize direct execution of high-impact remote changes without adequate safeguards.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The firewall example directly updates host packet-filtering rules and ends with a blanket DROP for port 443 traffic not matching fetched CDN IPs. Without prominent interruption warnings and rollback instructions, users may sever valid traffic paths, break health checks, or lock themselves out of services.

External Transmission

Medium

Category: Data Exfiltration
Content: ZONE_ID="你的ZONE_ID" # ... # 1. 开启Always Use HTTPS curl -s -X PATCH "https://api.cloudflare.com/client/v4/zones/${ZONE_ID}/settings/always_use_https" \ -H "Authorization: Bearer ${API_TOKEN}" \ -H "Content-Type: application/json" \ --data '{"value":"on"}' | jq '.success'
Confidence: 89% confidence
Finding: https://api.cloudflare.com/

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.