Security audit

Dk Helloworld 001

Security checks for vulnerabilities and agentic risk

Overview

This is a minimal text echo skill with confusing unrelated API notes in its documentation, but no implemented network, credential, publish, delete, file, or persistence behavior.

Safe to install as a simple echo/demo skill, but do not send secrets or sensitive text because echoed input may appear in logs or UI output. The publisher should remove the unrelated registry API table from SKILL.md to avoid user and reviewer confusion.

SkillSpector

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (2)

Intent-Code Divergence

High

Confidence: 97% confidence
Finding: The file claims to be a simple 'hello world' printer, but then documents a substantially different authenticated registry API with search, download, publish, delete, and auth operations. This mismatch is dangerous because it obscures the true scope and capabilities of the skill, preventing reviewers and users from accurately assessing what the skill does and potentially hiding privileged or network-enabled behavior behind misleading documentation.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The skill description is overly broad ('print any text') and does not define when the skill should activate, what inputs are expected, or any limits on behavior. In agent environments, vague activation boundaries can cause unintended invocation, prompt collisions, or misuse in contexts where arbitrary user-controlled content should not be handled without constraints.

Static analysis

No suspicious patterns detected.