Security audit

Drop Files

Security checks for vulnerabilities and agentic risk

Overview

This skill does what it says: uploads HTML or Markdown to a public DropFiles link, with clear enough disclosure that links are public unless password-protected.

Install only if you are comfortable sending selected HTML or Markdown to DropFiles for public link sharing. Do not use it for confidential, regulated, credential-bearing, or internal-only content unless you intentionally choose password protection and accept third-party hosting.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The invocation guidance is broad enough to trigger on generic requests to share, publish, or generate links for content, which increases the chance an agent will upload material outside a narrowly intended scope. In this skill’s context, that broad trigger directly causes transmission of user/model-produced content to a third-party public hosting service, creating a meaningful risk of accidental disclosure.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The description markets the skill as a simple way to get a shareable public link but does not prominently warn that content is sent to an external service and may become publicly accessible. That omission is dangerous because users may reasonably assume local processing or underestimate that sensitive prompts, reports, or generated artifacts will leave the trusted environment.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The skill’s purpose is to upload user-generated content to a third-party public service, but the invocation description does not prominently warn that content will be sent off-platform and may become publicly accessible by default. This creates a real risk of unintended disclosure of sensitive data because users may invoke the skill expecting simple sharing behavior without understanding the privacy implications.

Ssd 3

Medium

Confidence: 93% confidence
Finding: The skill explicitly encourages publishing arbitrary generated content via a public URL, which is a common path for leaking sensitive user data, hidden model context, or internal analysis embedded in generated outputs. Because the trigger is framed in ordinary language workflows, an agent may apply it automatically without recognizing that the content contains confidential or unintended information.

Static analysis

No suspicious patterns detected.