Back to skill

Security audit

skill-vetter

Security checks for vulnerabilities and agentic risk

Overview

This appears to be a skill-review helper, but it is advertised too broadly while requesting command execution and file-writing powers.

Review before installing. It is not clearly malicious, but it should be narrowed to pre-install skill security review, preferably with read-only defaults and explicit consent for any command execution, network lookup, callback use, or saved report output.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
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 (10)

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The manifest markets the skill as a broad efficiency/automation tool, while the body defines a specialized security-vetting protocol. This mismatch can cause the skill to be invoked in unrelated contexts, giving a review-oriented skill access to read/exec/write capabilities under misleading pretenses and increasing the chance of unsafe or unintended use.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

Overly broad invocation language makes the skill eligible for many generic productivity requests unrelated to security review. Because the skill also has powerful tools, this increases the probability of accidental activation in contexts where users did not intend to grant a vetting/audit tool read/exec/write behavior.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 136)May include surrounding context.

md
# "SKILL.md"

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The primary displayName, summary, and description are written in Chinese, while no language preference, opt-in, or region-specific justification is provided. This creates a locale-policy concern because the skill appears to impose a language on users rather than offering a choice.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

Broad command execution is excessive for a skill whose stated purpose is to inspect skill files and assess risk. If invoked on untrusted content, exec capability can turn a static review process into active code execution or shell-driven data access, creating avoidable host and privacy risk.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

File-write capability is not inherent to producing an audit assessment and creates unnecessary modification risk. In the context of a skill that may inspect untrusted packages, write access could alter workspace files, create misleading artifacts, or support unwanted installation-related behavior.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The documentation says the skill should review other skills before installation, but it also declares write and exec capabilities that can produce side effects during that review. For a security-vetting workflow, this expands the blast radius from passive analysis into potentially modifying files or running commands on the host, which is unnecessary for many reviews and risky if triggered automatically.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

Claiming compatibility with virtually any agent platform without narrowing when the skill should trigger broadens deployment and invocation surface. In practice, this can cause the skill to be imported into environments with different trust models, tool semantics, or weaker safeguards than assumed by the document.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · SKILL.md (reported line 141)May include surrounding context.

"(请参考skill目录中的脚本文件)"

...

3. 读取 SKILL.md 内容

curl -s "https://raw.githubusercontent.com/example/skill-repo/main/skills/my-skill/SKILL.md"

text

```text

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The markdown describes capabilities involving file writes, external API access, and command execution but does not provide a consolidated, explicit warning about privacy, network egress, and system-impacting behavior. Users may therefore enable or invoke the skill without understanding that review actions could expose metadata externally or alter local state.

Content

No source excerpt is available for this finding.