ClawHub
Back to skill

Security audit

Web Notepad

Security checks across malware telemetry and agentic risk

Overview

This skill is mostly a form-management guide, but mixed security-review claims and examples for live API changes, webhooks, exports, and permission changes need human review before installation.

Install only if you intend to let the agent help operate a Web Notepad-style form service. Treat all curl examples as live administrative actions: verify the API host, use least-privilege keys, prefer a sandbox first, and require explicit approval before creating forms, importing/exporting submissions, changing RBAC assignments, or enabling webhooks.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (12)

Intent-Code Divergence

High
Confidence
95% confidence
Finding
The skill metadata and description present it as a form-management tool, but the activation text and semantics also claim security detection, compliance auditing, and vulnerability scanning use cases. This mismatch can cause an agent to invoke the skill in the wrong context and produce security-review style outputs while actually performing unrelated external API actions, increasing the risk of unsafe or unauthorized operations.

Intent-Code Divergence

Medium
Confidence
92% confidence
Finding
The capability table advertises deep vulnerability scanning, CVE correlation, risk scoring, and threat-intelligence features that do not match the rest of the documented form-management behavior. In an agent setting, contradictory capability claims can mislead orchestration or users into trusting the skill for security-sensitive analysis it does not actually provide, while still exposing data to external endpoints.

Intent-Code Divergence

High
Confidence
96% confidence
Finding
The documented input/output contract models a scoring and review engine with grades and compliance findings, which conflicts with the surrounding examples of CRUD, webhook, and submission-management APIs. This creates a dangerous interface deception where an agent may send arbitrary content for 'review' yet trigger or rely on a tool that actually manages external form data, leading to misuse, incorrect trust decisions, or unintended data handling.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The description says to use the skill for broad security detection, compliance audit, vulnerability scanning, and encryption protection tasks, which is overly generic and not tightly scoped to the documented product behavior. Broad activation language can cause accidental invocation on unrelated sensitive tasks, sending data to an external service or prompting unsafe tool use under false assumptions about capability.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The examples include POST requests that create templates, forms, role assignments, webhooks, batch imports, and exports, all of which can modify live systems or transmit potentially sensitive submissions to external endpoints. Without prominent warnings, dry-run guidance, or explicit user confirmation requirements, an agent could follow these examples and cause unintended data changes or disclosures.

External Transmission

Medium
Category
Data Exfiltration
Content
# 1. 创建审批模板(可复用)
curl -X POST -H "Authorization: Bearer $WEB_NOTEPAD_API_KEY" \
  -H "Content-Type: application/json" \
  "https://api.web-notepad.example/v1/templates" \
  -d '{
    "name": "标准审批单模板",
    "fields": [
Confidence
87% confidence
Finding
https://api.web-notepad.example/

External Transmission

Medium
Category
Data Exfiltration
Content
# 2. 基于模板批量创建审批表单
curl -X POST -H "Authorization: Bearer $WEB_NOTEPAD_API_KEY" \
  -H "Content-Type: application/json" \
  "https://api.web-notepad.example/v1/forms/batch" \
  -d '{
    "templateId": "tpl_详情见说明",
    "items": [
Confidence
87% confidence
Finding
https://api.web-notepad.example/

External Transmission

Medium
Category
Data Exfiltration
Content
# 3. 配置Webhook推送审批事件
curl -X POST -H "Authorization: Bearer $WEB_NOTEPAD_API_KEY" \
  -H "Content-Type: application/json" \
  "https://api.web-notepad.example/v1/forms/{formId}/webhooks" \
  -d '{
    "url": "https://your-app.example/webhooks/approval",
    "events": ["submission.created", "submission.updated"],
Confidence
91% confidence
Finding
https://api.web-notepad.example/

External Transmission

Medium
Category
Data Exfiltration
Content
# 1. 创建部门角色
curl -X POST -H "Authorization: Bearer $WEB_NOTEPAD_API_KEY" \
  -H "Content-Type: application/json" \
  "https://api.web-notepad.example/v1/roles" \
  -d '{
    "name": "销售部提交员",
    "base": "viewer",
Confidence
85% confidence
Finding
https://api.web-notepad.example/

External Transmission

Medium
Category
Data Exfiltration
Content
# 2. 分配用户到角色
curl -X POST -H "Authorization: Bearer $WEB_NOTEPAD_API_KEY" \
  -H "Content-Type: application/json" \
  "https://api.web-notepad.example/v1/assignments" \
  -d '{
    "principalType": "user",
    "principalId": "user_详情见说明",
Confidence
90% confidence
Finding
https://api.web-notepad.example/

External Transmission

Medium
Category
Data Exfiltration
Content
# 1. 批量导入历史样本数据
curl -X POST -H "Authorization: Bearer $WEB_NOTEPAD_API_KEY" \
  -H "Content-Type: application/json" \
  "https://api.web-notepad.example/v1/forms/{formId}/submissions/batch" \
  -d '{"submissions": [<数据从data.jsonl流式读取>]}'
# ...
# 2. 启用并行查询与缓存
Confidence
92% confidence
Finding
https://api.web-notepad.example/

External Transmission

Medium
Category
Data Exfiltration
Content
# 2. 启用并行查询与缓存
curl -X POST -H "Authorization: Bearer $WEB_NOTEPAD_API_KEY" \
  -H "Content-Type: application/json" \
  "https://api.web-notepad.example/v1/forms/{formId}/submissions/export" \
  -d '{
    "format": "csv",
    "filters": {"dateRange": {"start": "2026-01-01", "end": "2026-01-15"}},
Confidence
84% confidence
Finding
https://api.web-notepad.example/

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.