Security audit
Security checks for vulnerabilities and agentic risk
This is a coherent forklift-domain reference skill with disclosed web lookup and report-generation behavior, and I did not find evidence of hidden exfiltration, persistence, or destructive behavior.
Before installing, be aware that this skill may perform live web searches for current standards, sales, pricing-adjacent, and tender information, and tender-broadcast requests can create a persistent HTML report in /workspace. Use it for forklift-related tasks, and review generated reports before sharing because they may contain the query topic and public tender details.
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<!--
Copyright (c) 2026 杨鹏飞 / 微信公众号「叉车技术老炮」
Licensed under the MIT License - 完整协议见 ../LICENSE.md
-->
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<!--
Copyright (c) 2026 杨鹏飞 / 微信公众号「叉车技术老炮」
Licensed under the MIT License - 完整协议见 ../LICENSE.md
-->
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<!-- 背景 -->
<rect width="720" height="420" fill="#ffffff"/>
<!-- 标题 -->
<text x="24" y="36" font-size="18" font-weight="700" fill="#1a2a4a">
中国叉车品牌销量排行
</text>
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<!-- 背景 -->
<rect width="720" height="420" fill="#ffffff"/>
<!-- 标题 -->
<text x="24" y="36" font-size="18" font-weight="700" fill="#1a2a4a">
中国叉车品牌销量排行
</text>
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<!-- 背景 -->
<rect width="720" height="420" fill="#ffffff"/>
<!-- 标题 -->
<text x="24" y="36" font-size="18" font-weight="700" fill="#1a2a4a">
中国叉车品牌销量排行
</text>
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<rect x="566" y="22" width="14" height="14" fill="#ff6b35"/>
<text x="586" y="34" font-size="12" fill="#1a2a4a">国外品牌</text>
<!-- 轴 -->
<line x1="70" y1="70" x2="70" y2="330" stroke="#d1d5db"/>
<line x1="70" y1="330" x2="680" y2="330" stroke="#d1d5db"/>
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<rect x="566" y="22" width="14" height="14" fill="#ff6b35"/>
<text x="586" y="34" font-size="12" fill="#1a2a4a">国外品牌</text>
<!-- 轴 -->
<line x1="70" y1="70" x2="70" y2="330" stroke="#d1d5db"/>
<line x1="70" y1="330" x2="680" y2="330" stroke="#d1d5db"/>
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<line x1="70" y1="70" x2="70" y2="310" stroke="#d1d5db"/>
<line x1="70" y1="310" x2="680" y2="310" stroke="#d1d5db"/>
<!-- 折线：x 均分 (680-70)/(n-1)，y = 310 - (值-下限)/(上限-下限)*240 -->
<polyline points="70,220 190,196 310,168 430,140 550,110 670,86"
fill="none" stroke="#ff6b35" stroke-width="3"
stroke-linejoin="round" stroke-linecap="round"/>
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
<line x1="70" y1="70" x2="70" y2="310" stroke="#d1d5db"/>
<line x1="70" y1="310" x2="680" y2="310" stroke="#d1d5db"/>
<!-- 折线:x 均分 (680-70)/(n-1),y = 310 - (值-下限)/(上限-下限)*240 -->
<polyline points="70,220 190,196 310,168 430,140 550,110 670,86"
fill="none" stroke="#ff6b35" stroke-width="3"
stroke-linejoin="round" stroke-linecap="round"/>
The README says the skill is triggered by a long list of common domain terms such as 'battery', 'BMS', 'charger', and brand names, without clarifying when those mentions should or should not invoke the skill. This creates ambiguous activation scope and increases the chance of unintended invocation during ordinary discussion of related equipment or components.
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.
rm -rf <skills 目录>/@user_bc886e82
各平台 skills 目录示例：Claude Code `~/.claude/skills/`、Cursor `~/.cursor/skills/`、
Codex `~/.codex/skills/`、Gemini CLI `~/.gemini/skills/`、CodeBuddy `~/.codebuddy/skills/`。
### 核心硬规则
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.
rm -rf <skills 目录>/@user_bc886e82
各平台 skills 目录示例:Claude Code `~/.claude/skills/`、Cursor `~/.cursor/skills/`、
Codex `~/.codex/skills/`、Gemini CLI `~/.gemini/skills/`、CodeBuddy `~/.codebuddy/skills/`。
### 核心硬规则
The English section lists many standalone terms like 'battery', 'charger', 'controller', and several brand names as trigger scenarios, but does not distinguish between relevant forklift-specific requests and ordinary mentions of those words. Because the full trigger list is deferred elsewhere, this README itself leaves activation behavior ambiguous.
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.
rm -rf /@user_bc886e82
Typical skills directories: Claude Code `~/.claude/skills/`, Cursor `~/.cursor/skills/`,
Codex `~/.codex/skills/`, Gemini CLI `~/.gemini/skills/`, CodeBuddy `~/.codebuddy/skills/`.
### Core Hard Rules
The skill states it will automatically match the output language to the user's input language, and later reinforces that no explicit user request is needed. This is a locale/language policy concern because it imposes language behavior by default rather than offering the user a choice or requiring opt-in.
The skill instructs the agent to create a file under /workspace and present it without requiring an explicit user-facing notice or confirmation at the time of file creation. File creation is a side effect on the local environment, and if normalized without disclosure it can surprise users, leak sensitive queried content into persistent artifacts, or be abused by prompt-injected workflows to generate unintended files.
These lines explicitly require automatic language matching for Chinese vs. English responses and state that the user does not need to request this behavior. That is a natural-language policy issue because the skill enforces a language choice without user opt-in.
Suspicious Unicode normalization or mixed-script content
The document's operational guidance, safety instructions, and troubleshooting steps are entirely presented in Chinese. Under the policy, forcing a specific language without user opt-in can be a natural-language policy violation unless the locale restriction is explicitly justified, which is not stated here.
The module states that all covered modules must automatically match the user's question language, making the locale behavior organization-wide and mandatory. This goes beyond a helpful default and becomes a forced policy unless the user explicitly switches languages.
These rules require the model to answer entirely in Chinese or English according to inferred input language, which is a language/locale policy constraint. Because the behavior is automatic rather than user-chosen, it can conflict with policies that require explicit language choice or opt-in.
This markdown file is natural-language content and appears to require Chinese comprehension throughout, including the title, instructions, and all knowledge entries. Under the policy rule, forcing a specific language without user opt-in is a violation unless the locale restriction is explicitly documented and justified, which is not present here.
Suspicious Unicode normalization or mixed-script content
Suspicious Unicode normalization or mixed-script content
The file instructs the model to always include a specific Chinese attribution string in responses. This imposes a language/locale requirement on output without user opt-in, and the document does not explain that the skill is limited to a Chinese-only or region-specific context.
No suspicious patterns detected.