Back to skill

Security audit

skill-quality-inspector

Security checks for vulnerabilities and agentic risk

Overview

This skill checks and organizes skill repositories, with repairs requiring your explicit authorization.

Select only repositories you intend to inspect, and review terminology, deprecation, and registry changes before authorizing repairs. Installation examples use unpinned sources with global installation and automatic confirmation; prefer reviewed versions and deliberate installation choices. Full organizational checks depend on separately reviewed skills-repo-admin tools and rules that are absent from this package.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (65)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
82% confidence
Finding

The declared purpose covers inspection and quality-related fixes, not skill installation. The supplied command performs an installation-related operation and contains no auditing behavior. Its location in a test README limits certainty: it could be an inert fixture, but no surrounding test logic establishes that supporting role.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Repository-writing instructions mandate Chinese without user opt-in.

Medium
Category
Output Handling
Confidence
98% confidence
Finding

L34 requires body text to be written in Chinese, allowing English only for technical terms. It offers no language choice or opt-in and provides no locale-specific justification for imposing this requirement.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding

Although this is not an MCP server configuration, the installation command executes an unpinned npm CLI and installs skill content from a mutable repository reference. A compromised future package or repository revision could introduce arbitrary code or malicious agent instructions; global installation and automatic confirmation increase exposure.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding

This command executes an unpinned npm CLI and installs repository-sourced agent instructions without an immutable source revision. The quality-auditor selector identifies a subskill rather than a reviewed version, leaving supply-chain exposure despite the narrower installation scope; this is not an MCP-specific issue.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: file_read.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

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 46)May include surrounding context.

md
| 任务 | 读取·调用(子技能) | grep 关键词 |
|------|---------------------|-------------|
| 技能质量检查 | [skills/quality-auditor/SKILL.md](skills/quality-auditor/SKILL.md) | 质量检查、门禁对账、定级 |
| 技能矛盾点修复 | [skills/contradiction-fixer/SKILL.md](skills/contradiction-fixer/SKILL.md) | 矛盾修复、断链、数字不一致 |
| 技能风险评估 | [skills/risk-evaluator/SKILL.md](skills/risk-evaluator/SKILL.md) | 风险评估、安全、维护、合规 |
| 技能冗余治理 | [skills/redundancy-reducer/SKILL.md](skills/redundancy-reducer/SKILL.md) | 冗余治理、重名、重叠 |

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 46)May include surrounding context.

md
| 任务 | 读取·调用(子技能) | grep 关键词 |
|------|---------------------|-------------|
| 技能质量检查 | [skills/quality-auditor/SKILL.md](skills/quality-auditor/SKILL.md) | 质量检查、门禁对账、定级 |
| 技能矛盾点修复 | [skills/contradiction-fixer/SKILL.md](skills/contradiction-fixer/SKILL.md) | 矛盾修复、断链、数字不一致 |
| 技能风险评估 | [skills/risk-evaluator/SKILL.md](skills/risk-evaluator/SKILL.md) | 风险评估、安全、维护、合规 |
| 技能冗余治理 | [skills/redundancy-reducer/SKILL.md](skills/redundancy-reducer/SKILL.md) | 冗余治理、重名、重叠 |

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 · references/directory-standardization.md (reported line 52)May include surrounding context.

md
- [skills/directory-standardizer/SKILL.md](../skills/directory-standardizer/SKILL.md) — 本篇的执行落地层:本篇给"五层职责边界 + 命名约定 + Level 判定速查 + 错配清单",子技能给对齐目录、补路由、消孤儿的操作步骤(见其「工作流」)。
- 边界:**该不该改归本篇(判定),怎么改归子技能(执行)**。尤其 Level A/B 的取舍先读本篇第四节速查表,避免"为凑 A 造脚本"。
- 定级联动:结构分档的事实来源始终是 `skills-repo-admin` 的 `audit_architecture.py`;本仓 [skills/quality-auditor/SKILL.md](../skills/quality-auditor/SKILL.md) 只在其之上做跨文件一致性补充,不重复判分。

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 · references/directory-standardization.md (reported line 52)May include surrounding context.

md
- [skills/directory-standardizer/SKILL.md](../skills/directory-standardizer/SKILL.md) — 本篇的执行落地层:本篇给"五层职责边界 + 命名约定 + Level 判定速查 + 错配清单",子技能给对齐目录、补路由、消孤儿的操作步骤(见其「工作流」)。
- 边界:**该不该改归本篇(判定),怎么改归子技能(执行)**。尤其 Level A/B 的取舍先读本篇第四节速查表,避免"为凑 A 造脚本"。
- 定级联动:结构分档的事实来源始终是 `skills-repo-admin` 的 `audit_architecture.py`;本仓 [skills/quality-auditor/SKILL.md](../skills/quality-auditor/SKILL.md) 只在其之上做跨文件一致性补充,不重复判分。

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 · references/quality-dimensions.md (reported line 65)May include surrounding context.

md
## 相关子技能与层次边界(L2→L3)

- [skills/quality-auditor/SKILL.md](../skills/quality-auditor/SKILL.md) — 本篇的执行落地层:本篇给"门禁→动作映射 + 脚本 vs 人工分工 + 判定口径 + 坑",子技能给一次完整质检的执行步骤与报告产出(见其「工作流」与「限制」)。
- 边界:**口径归本篇,跑批归子技能**。写任何门禁正则前必读本篇第三点五节(规范文本是唯一依据,不能拿现有仓库写法反推标准)。
- 分工联动:结构类失配交 [skills/directory-standardizer/SKILL.md](../skills/directory-standardizer/SKILL.md);跨文件矛盾与断链交 [skills/contradiction-fixer/SKILL.md](../skills/contradiction-fixer/SKILL.md);本篇清单只负责"查出来并归类"。

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 · references/quality-dimensions.md (reported line 65)May include surrounding context.

md
## 相关子技能与层次边界(L2→L3)

- [skills/quality-auditor/SKILL.md](../skills/quality-auditor/SKILL.md) — 本篇的执行落地层:本篇给"门禁→动作映射 + 脚本 vs 人工分工 + 判定口径 + 坑",子技能给一次完整质检的执行步骤与报告产出(见其「工作流」与「限制」)。
- 边界:**口径归本篇,跑批归子技能**。写任何门禁正则前必读本篇第三点五节(规范文本是唯一依据,不能拿现有仓库写法反推标准)。
- 分工联动:结构类失配交 [skills/directory-standardizer/SKILL.md](../skills/directory-standardizer/SKILL.md);跨文件矛盾与断链交 [skills/contradiction-fixer/SKILL.md](../skills/contradiction-fixer/SKILL.md);本篇清单只负责"查出来并归类"。

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 · skills/quality-auditor/SKILL.md (reported line 7)May include surrounding context.

md
source:
  type: original
  repo: skills-repo/skill-quality-inspector
  path: skills/quality-auditor/SKILL.md
  version: "1.0.0"
  updated: "2026-08-14"
metadata:

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 47)May include surrounding context.

md
| 任务 | 读取·调用(子技能) | grep 关键词 |
|------|---------------------|-------------|
| 技能质量检查 | [skills/quality-auditor/SKILL.md](skills/quality-auditor/SKILL.md) | 质量检查、门禁对账、定级 |
| 技能矛盾点修复 | [skills/contradiction-fixer/SKILL.md](skills/contradiction-fixer/SKILL.md) | 矛盾修复、断链、数字不一致 |
| 技能风险评估 | [skills/risk-evaluator/SKILL.md](skills/risk-evaluator/SKILL.md) | 风险评估、安全、维护、合规 |
| 技能冗余治理 | [skills/redundancy-reducer/SKILL.md](skills/redundancy-reducer/SKILL.md) | 冗余治理、重名、重叠 |
| 技能目录规范化 | [skills/directory-standardizer/SKILL.md](skills/directory-standardizer/SKILL.md) | 目录规范、五层、路由孤儿 |

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 47)May include surrounding context.

md
| 任务 | 读取·调用(子技能) | grep 关键词 |
|------|---------------------|-------------|
| 技能质量检查 | [skills/quality-auditor/SKILL.md](skills/quality-auditor/SKILL.md) | 质量检查、门禁对账、定级 |
| 技能矛盾点修复 | [skills/contradiction-fixer/SKILL.md](skills/contradiction-fixer/SKILL.md) | 矛盾修复、断链、数字不一致 |
| 技能风险评估 | [skills/risk-evaluator/SKILL.md](skills/risk-evaluator/SKILL.md) | 风险评估、安全、维护、合规 |
| 技能冗余治理 | [skills/redundancy-reducer/SKILL.md](skills/redundancy-reducer/SKILL.md) | 冗余治理、重名、重叠 |
| 技能目录规范化 | [skills/directory-standardizer/SKILL.md](skills/directory-standardizer/SKILL.md) | 目录规范、五层、路由孤儿 |

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 · references/contradiction-detection.md (reported line 47)May include surrounding context.

md
## 相关子技能与层次边界(L2→L3)

- [skills/contradiction-fixer/SKILL.md](../skills/contradiction-fixer/SKILL.md) — 本篇的执行落地层:本篇给"矛盾类型学 + 修复 SOP + 坑"的判断依据,子技能给可直接调用的操作步骤(见其「工作流」与「限制」两节)。
- 边界:**定性归本篇,动手归子技能**。判断一条 finding 是阻断还是警告、术语选哪个写法为准,读本篇;实际执行扫描 → 定级 → 改文件 → 回归,走子技能。
- 风险联动:若矛盾被判为 P1/P2 需要排期,交 [skills/risk-evaluator/SKILL.md](../skills/risk-evaluator/SKILL.md) 定级,不在修复流程里自行拍板 SLA。

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 · references/contradiction-detection.md (reported line 47)May include surrounding context.

md
## 相关子技能与层次边界(L2→L3)

- [skills/contradiction-fixer/SKILL.md](../skills/contradiction-fixer/SKILL.md) — 本篇的执行落地层:本篇给"矛盾类型学 + 修复 SOP + 坑"的判断依据,子技能给可直接调用的操作步骤(见其「工作流」与「限制」两节)。
- 边界:**定性归本篇,动手归子技能**。判断一条 finding 是阻断还是警告、术语选哪个写法为准,读本篇;实际执行扫描 → 定级 → 改文件 → 回归,走子技能。
- 风险联动:若矛盾被判为 P1/P2 需要排期,交 [skills/risk-evaluator/SKILL.md](../skills/risk-evaluator/SKILL.md) 定级,不在修复流程里自行拍板 SLA。

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 · references/quality-dimensions.md (reported line 67)May include surrounding context.

md
- [skills/quality-auditor/SKILL.md](../skills/quality-auditor/SKILL.md) — 本篇的执行落地层:本篇给"门禁→动作映射 + 脚本 vs 人工分工 + 判定口径 + 坑",子技能给一次完整质检的执行步骤与报告产出(见其「工作流」与「限制」)。
- 边界:**口径归本篇,跑批归子技能**。写任何门禁正则前必读本篇第三点五节(规范文本是唯一依据,不能拿现有仓库写法反推标准)。
- 分工联动:结构类失配交 [skills/directory-standardizer/SKILL.md](../skills/directory-standardizer/SKILL.md);跨文件矛盾与断链交 [skills/contradiction-fixer/SKILL.md](../skills/contradiction-fixer/SKILL.md);本篇清单只负责"查出来并归类"。

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 · references/quality-dimensions.md (reported line 67)May include surrounding context.

md
- [skills/quality-auditor/SKILL.md](../skills/quality-auditor/SKILL.md) — 本篇的执行落地层:本篇给"门禁→动作映射 + 脚本 vs 人工分工 + 判定口径 + 坑",子技能给一次完整质检的执行步骤与报告产出(见其「工作流」与「限制」)。
- 边界:**口径归本篇,跑批归子技能**。写任何门禁正则前必读本篇第三点五节(规范文本是唯一依据,不能拿现有仓库写法反推标准)。
- 分工联动:结构类失配交 [skills/directory-standardizer/SKILL.md](../skills/directory-standardizer/SKILL.md);跨文件矛盾与断链交 [skills/contradiction-fixer/SKILL.md](../skills/contradiction-fixer/SKILL.md);本篇清单只负责"查出来并归类"。

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 · references/risk-assessment.md (reported line 49)May include surrounding context.

md
- [skills/risk-evaluator/SKILL.md](../skills/risk-evaluator/SKILL.md) — 本篇的执行落地层:本篇给"三维风险 + 三级定级 + 定级矩阵 + 处置流程 + 坑",子技能给一次完整风险评估的执行步骤与矩阵产出(见其「工作流」与「限制」)。
- 边界:**定级标准归本篇,采信号与出矩阵归子技能**。安全启发式只给"可疑信号",是否定罪需按本篇第五节结合上下文判断,误报同样算 bug。
- 处置联动:定级后的实际修复交 [skills/contradiction-fixer/SKILL.md](../skills/contradiction-fixer/SKILL.md)(内容矛盾类)或仓库负责人;本层严格保持"只定级、不改文件"的只读边界。

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 · references/risk-assessment.md (reported line 49)May include surrounding context.

md
- [skills/risk-evaluator/SKILL.md](../skills/risk-evaluator/SKILL.md) — 本篇的执行落地层:本篇给"三维风险 + 三级定级 + 定级矩阵 + 处置流程 + 坑",子技能给一次完整风险评估的执行步骤与矩阵产出(见其「工作流」与「限制」)。
- 边界:**定级标准归本篇,采信号与出矩阵归子技能**。安全启发式只给"可疑信号",是否定罪需按本篇第五节结合上下文判断,误报同样算 bug。
- 处置联动:定级后的实际修复交 [skills/contradiction-fixer/SKILL.md](../skills/contradiction-fixer/SKILL.md)(内容矛盾类)或仓库负责人;本层严格保持"只定级、不改文件"的只读边界。

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 · skills/contradiction-fixer/SKILL.md (reported line 7)May include surrounding context.

md
source:
  type: original
  repo: skills-repo/skill-quality-inspector
  path: skills/contradiction-fixer/SKILL.md
  version: "1.0.0"
  updated: "2026-08-14"
metadata:

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 48)May include surrounding context.

md
|------|---------------------|-------------|
| 技能质量检查 | [skills/quality-auditor/SKILL.md](skills/quality-auditor/SKILL.md) | 质量检查、门禁对账、定级 |
| 技能矛盾点修复 | [skills/contradiction-fixer/SKILL.md](skills/contradiction-fixer/SKILL.md) | 矛盾修复、断链、数字不一致 |
| 技能风险评估 | [skills/risk-evaluator/SKILL.md](skills/risk-evaluator/SKILL.md) | 风险评估、安全、维护、合规 |
| 技能冗余治理 | [skills/redundancy-reducer/SKILL.md](skills/redundancy-reducer/SKILL.md) | 冗余治理、重名、重叠 |
| 技能目录规范化 | [skills/directory-standardizer/SKILL.md](skills/directory-standardizer/SKILL.md) | 目录规范、五层、路由孤儿 |

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 48)May include surrounding context.

md
|------|---------------------|-------------|
| 技能质量检查 | [skills/quality-auditor/SKILL.md](skills/quality-auditor/SKILL.md) | 质量检查、门禁对账、定级 |
| 技能矛盾点修复 | [skills/contradiction-fixer/SKILL.md](skills/contradiction-fixer/SKILL.md) | 矛盾修复、断链、数字不一致 |
| 技能风险评估 | [skills/risk-evaluator/SKILL.md](skills/risk-evaluator/SKILL.md) | 风险评估、安全、维护、合规 |
| 技能冗余治理 | [skills/redundancy-reducer/SKILL.md](skills/redundancy-reducer/SKILL.md) | 冗余治理、重名、重叠 |
| 技能目录规范化 | [skills/directory-standardizer/SKILL.md](skills/directory-standardizer/SKILL.md) | 目录规范、五层、路由孤儿 |

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 · references/contradiction-detection.md (reported line 49)May include surrounding context.

md
- [skills/contradiction-fixer/SKILL.md](../skills/contradiction-fixer/SKILL.md) — 本篇的执行落地层:本篇给"矛盾类型学 + 修复 SOP + 坑"的判断依据,子技能给可直接调用的操作步骤(见其「工作流」与「限制」两节)。
- 边界:**定性归本篇,动手归子技能**。判断一条 finding 是阻断还是警告、术语选哪个写法为准,读本篇;实际执行扫描 → 定级 → 改文件 → 回归,走子技能。
- 风险联动:若矛盾被判为 P1/P2 需要排期,交 [skills/risk-evaluator/SKILL.md](../skills/risk-evaluator/SKILL.md) 定级,不在修复流程里自行拍板 SLA。