Back to skill

Security audit

职场高效能习惯养成教练

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed Markdown-only workplace coaching skill with broad but purpose-aligned coaching prompts and no executable or data-exfiltration behavior.

Before installing, be aware that the skill is mainly Chinese-language coaching content and may engage on broad workplace-problem descriptions. Treat relationship audits, mission statements, and check-in logs as private notes, and do not record or identify other people without consent and appropriate workplace/legal permission.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (24)

Context Leakage

High
Category
Data Exfiltration
Confidence
85% confidence
Finding

The example task 'record the conversation and the other person's reaction' can encourage users to capture interpersonal interactions in ways that may include sensitive workplace information, private data, or audio/transcript content from third parties. In a coaching skill context, this is more risky because users may apply the instruction to real colleagues without considering consent, employer policy, or data minimization.

Content

Scanner excerpt · README.md (reported line 61)May include surrounding context.

md
#### 📏 Every Check-in Task Is Verifiable

Most habit plans fail because "be more proactive today" isn't checkable. Here each daily task carries an explicit completion standard — "log at least 3 language substitutions", "produce a list tagged with all four quadrants", "record the conversation and the other person's reaction". Done or not done is unambiguous, and each task is capped at 15-30 minutes.

#### 🎯 Diagnosis Covers All Seven, No Cherry-Picking

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · index.html (reported line 148)May include surrounding context.

html
</head>
<body>

<!-- Nav -->
<nav><div class="nav-inner">
<div class="nav-logo">🧭 mu-highly-effective · <span data-en="Effective Habits Coach" data-zh="职场高效能习惯养成教练">职场高效能习惯养成教练</span></div>
<ul class="nav-links">

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · index.html (reported line 225)May include surrounding context.

html
</div>
</div></section>

<!-- ✨ 核心亮点 -->
<section id="highlights" style="background:var(--bg-warm)"><div class="container">
<h2 class="section-title"><span data-en="✨ Core <em>Highlights</em>" data-zh="✨ <em>核心</em>亮点">✨ <em>核心</em>亮点</span></h2>
<p class="section-desc"><span data-en="Not another book summary &mdash; a habit operating system you can actually run." data-zh="不是又一个读书笔记,而是一套可以直接跑起来的习惯操作系统。">不是又一个读书笔记,而是一套可以直接跑起来的习惯操作系统。</span></p>

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · index.html (reported line 342)May include surrounding context.

html
</div>
</div></section>

<!-- 🚀 工作流 -->
<section id="workflows" style="background:var(--bg-warm)"><div class="container">
<h2 class="section-title"><span data-en="🚀 <em>Workflows</em>" data-zh="🚀 <em>工作流</em>">🚀 <em>工作流</em></span></h2>
<p class="section-desc"><span data-en="6 independent workflows. Trigger and run on demand." data-zh="6 个独立工作流,触发即跑,互不依赖。">6 个独立工作流,触发即跑,互不依赖。</span></p>

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The workflow trigger says it activates on 'Any pasted workplace situation, conflict, or confusion,' which is extremely broad and overlaps with ordinary conversation about work. The description does not provide exclusion conditions or negative examples, so it is unclear when the skill should not activate.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The mode is triggered when the user pastes 'a workplace situation, problem, or question directly,' which could match a very large share of normal work-related chat. Without stricter boundaries or opt-in phrasing, the skill may activate unintentionally.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
85% 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 · README.md (reported line 143)May include surrounding context.

git clone https://github.com/muippt/mu-highly-effective.git ~/.claude/skills/mu-highly-effective

text

> Using a different agent? Just drop the folder wherever your tool loads skills from. Project-level works too: `.claude/skills/mu-highly-effective`.

**2. Verify** — restart your agent and confirm the skill is picked up

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The README states that users can paste 'any workplace situation, conflict, or confusion' and the skill will auto-identify governing habits. 'Any' makes the trigger scope too expansive and does not clarify limits, increasing the chance of unintended invocation from ordinary workplace discussion.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The workflow trigger phrases are extremely broad, such as activating on essentially any pasted workplace scenario or confusion. In an agent environment, this can cause unintended skill invocation during ordinary conversation, which may override user intent, inject unsolicited behavioral framing, or interfere with other skills competing on similar natural-language triggers.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The '快捷模式' trigger is defined so broadly that nearly any work-related problem statement could activate it. This increases the chance of accidental routing, causing the agent to apply this skill's structured coaching output when the user may have intended a different task or a more general response.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The usage guidance says that pasting any workplace scenario, conflict, or confusion will automatically run the skill. This is effectively a collision-prone activation rule that may capture normal conversation and trigger behavior without informed user selection, which is a prompt-routing and reliability risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

L003 的显示名及全文指令均以中文固定表述该技能的交互方式,且文档未提供用户可选择回复语言或语言切换的说明。按组织语言/locale 政策标准,若技能默认并隐含强制单一语言而无用户选择或明确地域限定,可能构成自然语言策略违规。

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

L027 将快捷模式触发定义为“用户直接贴出职场场景、困惑、问题”,这几乎覆盖了常见求助表达,但没有给出边界、限定上下文或排除条件。这样的描述容易让技能在一般性职场聊天中被过度触发,属于模糊且范围过大的自然语言触发条件。

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

L050 使用“用户要求做习惯诊断 / 写使命宣言 / 定制打卡计划 / 盘点情感账户 / 整理时间矩阵”作为触发,但未说明这些要求需要多明确、是否接受近义表达、以及哪些相似请求不应触发。对 markdown 技能说明而言,这种缺乏范围约束的触发表述仍属模糊触发。

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

L056 以“打卡”“21天”“49天”“7天计划”等作为触发,其中“打卡”和“21天”等表述在日常对话中非常常见,并不天然限定在本技能领域。缺少与七个习惯、职场习惯养成等上下文绑定,容易与普通计划讨论发生碰撞。

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The workflow trigger text says users can 'Paste any situation' to invoke workplace Q&A, which is broad and can cause the skill to activate on generic work-related input rather than an explicit user request to use this skill. In an agent environment with multiple skills or routing logic, vague activation language increases the chance of unintended invocation and context capture.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The quick mode advertises activation when a user 'Paste[s] a workplace situation,' which is effectively an any-text trigger for a very common class of user input. This can lead to accidental skill engagement, misrouting, and unintended processing of sensitive workplace details when the user did not deliberately choose this skill.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This markdown file presents all user-facing content in Chinese from the title onward, with no indication that users may choose another language or that the skill is intended only for a Chinese-speaking audience. The policy explicitly calls out forced language or locale as a natural-language violation when no opt-in or justification is provided.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The file is entirely written in Chinese and prescribes Chinese-language prompts and output without offering the user a language choice. In a multilingual agent environment, this can exclude or confuse users, reduce informed consent around the interaction format, and create accessibility and usability risks, though it does not appear to enable code execution or data exfiltration.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This markdown file contains user-facing instructions and scenario responses exclusively in Chinese, with no indication that users may choose another language. Under the natural-language policy rule, forcing a specific language without user opt-in is a locale-policy violation unless clearly documented and justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

L129 标注“语言 | 中文内容,英文元数据”,体现出技能默认语言被预设为中文,但未见用户可选择输出语言或显式 opt-in 说明。按语言/locale 政策,若技能强制特定语言而不给用户选择,属于自然语言层面的合规风险。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

L268 明确规定“用‘你’不用‘我们’”,属于对输出语言风格的强制要求,但未说明可根据用户偏好、文化语境或礼貌层级调整。虽然主要是风格约束,但它仍体现了未给用户语言表达选择的固定化策略。

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

The page initializes the interface in Chinese by default via lang='zh' and applyLang('zh'), which imposes a language choice before any user selection. Although a language toggle exists, the default still forces a specific locale without explicit opt-in or preference detection.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

This markdown file presents all instructions and content in Chinese, and nowhere indicates that the user can choose another language or that the skill is intentionally limited to a Chinese-speaking audience. Under the language/locale policy, forcing a specific language without user opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.