Back to skill

Security audit

Ui Ux Design Expert

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because it found unsafe guidance about saving form data and reporting error details, along with instructions to change persistent knowledge files; coverage was partial.

Consider not installing until the privacy safeguards and approval requirements are reviewed. The scan inspected 14 of 16 components, with two partially inspected, so its evidence is incomplete; the suspicious verdict reflects risky guidance, not a finding that the author acted maliciously.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 18)May include surrounding context.

md
| kb/typography.md | 法则(TY)、排版与字体:字体加载策略、字体栈与回退、中西文混排、行高/字距、流体排版(clamp 边界)、可变字体、数值排版(tabular-nums)、打印样式(@media print) | 字体选型/加载、中西文规则、流体排版/可变字体、排版细节、打印适配 |

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 48)May include surrounding context.

md
- 字体选型与加载 / 字体栈 / 中西文混排 / 行高字距 / 数字排版 / 打印样式 → `kb/typography.md`

YARA rule 'agent_skill_mcp_tool_poisoning_metadata': MCP/tool metadata poisoning indicators in tool schemas or skill manifests [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · kb/aria-patterns.md (reported line 81)May include surrounding context.

md
6. Carousel(轮播)
- 键盘:自动轮播在**任一元素获得键盘焦点时停止**,除非用户激活轮播控件不恢复;Tab/Shift+Tab 正常序列;激活上一张/下一张**不移动焦点**(便于反复激活);轮播控制按钮是轮播内 Tab 序列**第一元素**;slide 选择器用 tabs 时按 Tabs 模式。
- ARIA:容器 `role="region"` 或 `group` + `aria-roledescription="carousel"` + 可见标签 `aria-labelledby`(否则 `aria-label`;标签不含「carousel」一词);每 slide `role="group"` + `aria-roledescription="slide"` + accessible name(无名用「3 of 10」序号);停止/重启按钮标签随动作变化(Stop/Start slide rotation),因标签变化**不设** aria-pressed;Tabbed 变体 slide 容器 `role="tabpanel"`(去 roledescription)+ tablist 选择器;Grouped 变体选择按钮组 `role="group"`+`aria-label`,当前 slide 选择按钮 `aria-disabled="true"`([AP-14],优于 HTML disabled);slide 集合可�

Automatic form-draft storage is recommended without a privacy warning.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

G-58 broadly recommends automatically saving form drafts in localStorage without excluding sensitive fields or specifying retention and deletion controls. If applied to forms containing personal information or credentials, this creates persistent browser copies accessible to same-origin JavaScript, increasing exposure through XSS or shared-device access; this is unsafe implementation guidance rather than evidence of an existing data leak.

Content

No source excerpt is available for this finding.

Automatic error reporting transmits URLs and referrers without a privacy warning.

Medium
Category
Output Handling
Confidence
97% confidence
Finding

The error-page guidance instructs automatic reporting of the URL, error code, and referrer without requiring sensitive-data filtering. If implemented literally, URLs and available referrer values could expose tokens, personal information, or private resource identifiers to telemetry storage and its readers. Error monitoring is legitimate in this UI/UX context, and no malicious destination is specified, but the missing privacy safeguards make this an unsafe implementation recommendation rather than evidence of active exfiltration.

Content

No source excerpt is available for this finding.

The knowledge-feedback workflow modifies persistent knowledge-base files without a user-facing warning.

Low
Category
Output Handling
Confidence
86% confidence
Finding

The feedback workflow directs the agent to append entries to persistent knowledge files and update their metadata when it encounters new defects or outdated guidance, without requiring user approval. During an ordinary UI task, this can cause unexpected changes to the installed skill and allow unreviewed task-derived conclusions to influence future sessions; provenance labels help trace changes but do not authorize or validate them.

Content

No source excerpt is available for this finding.