Back to skill

Security audit

安全生产专家

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent Chinese workplace-safety standards advisor with disclosed local lookup scripts and no evidence of hidden execution, exfiltration, or destructive behavior.

Install this only if you want a China-focused safety-production compliance reference tool. Configure SAFETY_STANDARDS_ROOT and SAFETY_CORPUS only to directories you are comfortable letting the skill search, and verify any legal or safety-critical answer against official sources or qualified professionals.

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
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (20)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

声明描述的是一个面向最终用户的安全生产问答/顾问技能,核心能力应是理解用户咨询并基于法律法规、标准进行回答。实际代码仅是一个用于准备参考资料的 setup 脚本:扫描本地标准文件库,从文件名中提取标准号并建立索引。这更像技能的数据预处理/构建工具,而不是技能本身宣称的运行时能力。虽然该脚本与“标准资料”主题相关,属于支持性数据准备,但如果把这段代码视为该技能的实际行为,其主要目的与声明的顾问问答能力存在明显差异,因此应判定为描述与行为不匹配。

Content

No source excerpt is available for this finding.

Hidden Instructions

High
Category
Prompt Injection
Confidence
60% confidence
Finding

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

Content

Scanner excerpt · references/标准目录.csv (reported line 1)May include surrounding context.

text
标准号,标题,主题,格式,相对路径
AQ-2048,AQ-2048-2012,其他/通用,.pdf,新法规\AQ-2048-2012.pdf
AQ-2048,AQ-2048-2012,其他/通用,.rar,新法规\AQ-2048-2012.rar
AQ-2048,AQ-2048-2012煤气安全隔断技术规范,其他/通用,.pdf,新标准\AQ-2048-2012煤气安全隔断技术规范.pdf

Lp1

High
Category
MCP Least Privilege
Confidence
75% confidence
Finding

The skill uses 'env' capability that is not listed in its permissions. This may indicate deceptive intent or missing permission declarations.

Content

No source excerpt is available for this finding.

Lp1

High
Category
MCP Least Privilege
Confidence
75% confidence
Finding

The skill uses 'file_write' capability that is not listed in its permissions. This may indicate deceptive intent or missing permission declarations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The file title and operating instructions are entirely Chinese and direct the agent to answer safety-production questions according to this file's rules whenever such questions are asked. There is no opt-in, fallback, or explicit allowance for answering in the user's preferred language, which is a natural-language locale policy concern.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

This README presents the skill's purpose, usage examples, installation guidance, and safety disclaimers exclusively in Chinese. Under the language/locale policy, forcing a single language without user opt-in or an explicit documented regional justification is a natural-language policy issue.

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 67)May include surrounding context.

md
cp -r safety-standards-advisor ~/.workbuddy/skills/

# Claude Code(项目级)
mkdir -p .claude/skills && cp -r safety-standards-advisor .claude/skills/

# Hermes Agent
cp -r safety-standards-advisor ~/.hermes/skills/

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.md (reported line 67)May include surrounding context.

md
cp -r safety-standards-advisor ~/.workbuddy/skills/

# Claude Code(项目级)
mkdir -p .claude/skills && cp -r safety-standards-advisor .claude/skills/

# Hermes Agent
cp -r safety-standards-advisor ~/.hermes/skills/

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The manifest says to load the skill when the user asks 'any' workplace safety-production-related question, followed by a broad list of examples. This creates an ambiguous activation boundary because it does not define when the skill should not activate or provide negative examples, increasing the chance of unintended invocation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The description is written as a China-specific legal advisor based on national laws and standards, but it does not explicitly state that the skill is limited to Chinese regulatory contexts or require user opt-in for that locale. This can violate language/locale policy expectations if activated for users seeking safety guidance in other jurisdictions.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest describes a skill for answering safety-production questions grounded in safety laws, industry standards, and regulations such as 安全生产法、消防法 and GB/AQ/TSG/JGJ standards. This index also catalogs unrelated legal domains such as 刑法, 未成年人保护法, and broader product-quality or metrology laws, which expands the retrievable corpus beyond the stated safety advisory purpose.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

This markdown file is entirely written as Chinese-only operational guidance and does not indicate that users may interact in other languages or that responses should match user preference. Under the policy rule for language/locale, a skill that effectively enforces one language without opt-in or justification can violate organizational language-choice requirements.

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 presents all instructions, legal references, and checklist items exclusively in Chinese. Under the policy rule, forcing a specific language without user opt-in or a clearly documented regional justification is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The module documentation says binary .doc files and scanned PDFs will be flagged for caller handling, but extract_doc() returns scanned=False. This mismatch can cause downstream logic to treat unsupported .doc files as successfully processed text documents, suppressing warnings or alternate conversion flows and leading to missed safety/compliance content in a standards-advisor skill.

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’s user-facing description and usage examples are entirely in Chinese, and later command help/output strings in the script follow the same pattern. Under the policy, forcing a specific language without user opt-in or a documented justification is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The top-level metadata fields and their values are entirely in Chinese, such as "总数", "说明", and "用法", which imposes a specific language/locale on consumers of the file. There is no indication elsewhere in the file that alternative locales are supported or that Chinese-only usage is an intentional, documented regional constraint.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The manifest centers on workplace safety-production issues such as hazard inspection, special operations, equipment acceptance, accident response, PPE, and emergency management. Entries such as 清洁生产促进法, 环境保护法, 排污许可管理条例, and 排污费征收使用管理条例 indicate the skill may serve environmental compliance topics that are adjacent to, but broader than, the declared safety-production scope.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

SQP-3 applies to all file types and covers natural-language language/locale policy violations. This JSON file contains titles and topics exclusively in Chinese on L1, with no indication that the skill is region-specific or that users can opt into this locale, which can conflict with policies requiring language choice or explicit justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

This markdown file presents all instructions and compliance guidance solely in Chinese. Under the policy check for natural-language violations, forcing a specific language without user opt-in can be a locale-policy issue when no language choice or justification is provided.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

The file’s natural-language content is entirely Chinese, and there is no indication that users may choose another language or that the topic is intentionally restricted to a Chinese-speaking or China-specific audience. Under the policy, forcing a specific language without user opt-in can be a locale-policy violation.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.