Back to skill

Security audit

Skill Health Monitor

Security checks for vulnerabilities and agentic risk

Overview

This skill is a simple in-memory health scoring helper with documentation quality issues but no hidden access, persistence, network use, or destructive behavior.

Before installing, treat this as a lightweight scoring utility rather than a robust security monitor. Prefer explicit invocation, validate metric inputs yourself, and expect Chinese-only output unless the publisher adds locale support.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
src/health-monitor.js:140
Finding

Unvalidated Metrics Can Bypass Health Alerts

Content
View full analysis
a + b, 0) / scores.length; return average; } ``` The resulting score is consumed by each alert handler using the following unsafe fallback pattern: ```javascript const score = skillData.healthScore || 1.0; ``` ### Technical Analysis The implementation does not verify that the six supplied metrics are finite numbers in the documented range of `0` through `1`. Logical OR fallbacks treat valid zero values as missing. Consequently, a metric of `0`, which should indicate the worst possible result, is replaced with the healthy default value `0.70`. Malformed values can also bypass alert classification. For example, if a metric is `NaN`, the calculated average becomes `NaN`. Each handler then evaluates: ```javascript const score = skillData.healthScore || 1.0; ``` Because `NaN` is falsy, it is replaced with `1.0`, causing all alert thresholds to pass and producing an `ok` result. Non-numeric strings, infinities, negative numbers, and values above `1` are likewise accepted without validation and can corrupt or manipulate the calculated result. ### Attack Path 1. A caller invokes `checkSkillHealth()` with attacker-controlled or otherwise untrusted metrics. 2. The caller supplies zero values, `NaN`, malformed str ...[truncated 1493 chars]
Remediation
View remediation
{ const value = metrics[key]; if (!Number.isFinite(value) || value < 0 || value > 1) { throw new RangeError(`${key} must be a finite number between 0 and 1`); } return value; }); const average = scores.reduce((sum, value) => sum + value, 0) / scores.length; if (!Number.isFinite(average)) { throw new Error('Calculated health score is invalid'); } return average; } ``` The handlers should also avoid truthiness-based fallback behavior: ```javascript const score = skillData.healthScore; if (!Number.isFinite(score) || score < 0 || score > 1) { throw new RangeError('Health score must be a finite number between 0 and 1'); } ``` ]]>
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (8)

Natural-Language Policy Violations

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The entire skill documentation and all user-facing invocation examples are presented only in Chinese, with no indication that users may choose another language or locale. This can constitute a language/locale policy violation when the skill implicitly requires a specific language without opt-in or justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

This markdown file presents the skill log entirely in Chinese, including headings, task descriptions, and role labels, without any indication that language selection is optional or user-configurable. Under the policy rule for natural-language violations, a fixed language/locale without opt-in can exclude users and violates the requirement to offer choice or clearly justify the locale constraint.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The usage section presents multiple ambiguous invocation examples like '检查 [技能名] 的健康状态' and '查看技能健康告警记录' without explicit activation constraints, authentication expectations, or parameter validation rules. In systems that infer tool usage from user language, these examples increase the risk of accidental invocation, overbroad data access, or unintended batch operations triggered by casual conversation.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The phrase '检查所有技能的健康状态' is a broad natural-language trigger that could match ordinary user requests without a clear invocation boundary. In agent environments that auto-activate skills from conversational text, this can cause unintended skill execution, unexpected context capture, or workflow hijacking when a user is merely discussing monitoring rather than intentionally invoking the skill.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The maintainer string "王的奴隶" literally contains self-demeaning or servile wording in natural language. Even though this is metadata in a markdown file, it is still visible organizational text and can violate workplace language or professionalism standards.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

该 Markdown 文档从标题到所有调用示例均仅以中文呈现,没有提供用户语言选择、替代语言版本或说明该技能仅面向特定中文环境。按规则,若技能在自然语言层面隐含强制特定语言而无用户选择,属于语言/locale 政策风险。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This JavaScript file consistently uses Chinese for the skill name, inline documentation, status messages, and recommendations, with no indication that the skill is region-specific or that users can opt into another language. That creates a natural-language locale policy issue because the skill behavior appears to assume a fixed language without user choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The manifest description is written entirely in Chinese and presents the skill in a single language, with no indication that language choice is configurable or that the skill is intentionally limited to a Chinese-speaking audience. Under the policy rules, a language-specific presentation without user opt-in or documented justification can be a locale-policy issue.

Content

No source excerpt is available for this finding.