Back to skill

Security audit

lovart skill

Security checks for vulnerabilities and agentic risk

Overview

The skill is a disclosed Lovart.ai API helper whose main risk is that it sends user design prompts to a third-party service and has some broad trigger wording.

Install this only if you intend to use Lovart.ai and are comfortable sending design prompts, brand details, and related usage metadata to that service under your Lovart account. Avoid submitting confidential product, campaign, or branding material unless third-party processing is approved, and invoke Lovart explicitly when possible to avoid accidental use from generic design requests.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (24)

Tp4

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding

The code is related to the declared Lovart.ai integration and does perform AI design/image generation plus task-status polling, so the general domain matches. However, the declared description specifically claims support for image generation, image editing, video creation, and template workflows. In this code chunk, only design/image generation examples and task monitoring are present. There are no calls or logic for editing existing images, creating videos, or running template-based workflows. Because the description overstates the implemented capabilities in a material way, this should be flagged as a mismatch.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The natural-language documentation, instructions, trigger descriptions, and warnings are presented in Chinese, which can impose a language constraint on users without any stated opt-in or justification. The file does not indicate that the skill is region-specific or provide an alternative language option.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · README.md (reported line 24)May include surrounding context.

生成产品图片:

bash
curl -X POST https://api.lovart.ai/v1/design/generate \
  -H "Authorization: Bearer $LOVART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README says the skill triggers when users mention phrases such as "AI 设计生成", "生成产品图片", and "创建营销素材", which are generic requests rather than narrowly scoped invocations. Because the trigger list lacks exclusions or context constraints, the skill could activate on common design-related conversation that does not specifically intend to use Lovart.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

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.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

Several trigger phrases are generic enough to match broad user requests like image generation, marketing assets, or AI design without clearly requiring Lovart-specific intent. This can cause the skill to activate unexpectedly, leading to unintended third-party API use, confusion about which tool is being invoked, and possible unnecessary exposure of user content to the external service.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

Nearly all user-facing comments and command-line messages are in Chinese, including errors, usage, and status output. This imposes a specific language on users without any opt-in or indication that the skill is intentionally limited to a Chinese-speaking context, which matches the language/locale policy violation criteria.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · README.md (reported line 24)May include surrounding context.

md
fi

# API 基础 URL
API_BASE="https://api.lovart.ai/v1"

# ========================================
# 示例 1: 生成产品图片

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · examples.sh (reported line 15)May include surrounding context.

sh
fi

# API 基础 URL
API_BASE="https://api.lovart.ai/v1"

# ========================================
# 示例 1: 生成产品图片

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · skill.md (reported line 86)May include surrounding context.

md
fi

# API 基础 URL
API_BASE="https://api.lovart.ai/v1"

# ========================================
# 示例 1: 生成产品图片

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · skill.md (reported line 100)May include surrounding context.

md
fi

# API 基础 URL
API_BASE="https://api.lovart.ai/v1"

# ========================================
# 示例 1: 生成产品图片

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · skill.md (reported line 113)May include surrounding context.

md
fi

# API 基础 URL
API_BASE="https://api.lovart.ai/v1"

# ========================================
# 示例 1: 生成产品图片

External Transmission

Medium
Category
Data Exfiltration
Confidence
70% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · examples.sh (reported line 23)May include surrounding context.

sh
generate_product_image() {
  echo "📸 生成产品图片..."

  response=$(curl -s -X POST "$API_BASE/design/generate" \
    -H "Authorization: Bearer $LOVART_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
70% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · examples.sh (reported line 45)May include surrounding context.

sh
generate_product_image() {
  echo "📸 生成产品图片..."

  response=$(curl -s -X POST "$API_BASE/design/generate" \
    -H "Authorization: Bearer $LOVART_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
70% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · examples.sh (reported line 119)May include surrounding context.

sh
echo "✅ 完成! 下载链接: $result_url"

      # 可选: 自动下载
      # curl -O $result_url
      break
    elif [ "$status" = "failed" ]; then
      echo ""

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The markdown instructs that the required prompt must be a detailed English description, which imposes a specific language requirement on all users. This is a natural-language policy concern because the skill does not offer a language choice or explain why English is mandatory.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
91% confidence
Finding

The skill explicitly instructs the agent to send user-provided design content to an external service using curl and an API key. This creates a real data-transmission boundary: user prompts, potentially sensitive images or business materials, and account-linked usage metadata are sent off-platform to Lovart.ai, which is a security/privacy concern if users are not clearly informed or if transmission is automatic.

Content

Scanner excerpt · prompt.md (reported line 35)May include surrounding context.

  • template_id: 模板 ID(如果适用)

第三步:调用 API

使用以下 curl 命令模板:

bash
curl -X POST https://api.lovart.ai/v1/design/generate \

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

This finding points to the concrete external endpoint used for design generation, confirming that the skill is designed to transmit data to a third-party domain. In context, this is more than incidental documentation because the skill operationalizes external requests and encourages routine submission of user content to that endpoint.

Content

Scanner excerpt · prompt.md (reported line 38)May include surrounding context.

使用以下 curl 命令模板:

bash
curl -X POST https://api.lovart.ai/v1/design/generate \
  -H "Authorization: Bearer $LOVART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
88% confidence
Finding

The example scenario includes a ready-to-use external API call for product-image generation, reinforcing that the skill will transmit user requests and derived prompts to Lovart.ai. Example code can normalize unsafe operational behavior by encouraging direct shell/API invocation without guardrails such as consent, validation, or content sensitivity checks.

Content

Scanner excerpt · prompt.md (reported line 119)May include surrounding context.

你的执行:

bash
curl -X POST https://api.lovart.ai/v1/design/generate \
  -H "Authorization: Bearer $LOVART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
88% confidence
Finding

This social-media example again demonstrates outbound transmission to a third-party API and could include marketing campaign details that organizations consider sensitive before launch. The repeated operational examples make the external transfer pattern part of the intended behavior, increasing the likelihood of routine data disclosure without sufficient friction or notice.

Content

Scanner excerpt · prompt.md (reported line 135)May include surrounding context.

你的执行:

bash
curl -X POST https://api.lovart.ai/v1/design/generate \
  -H "Authorization: Bearer $LOVART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

External Transmission

Medium
Category
Data Exfiltration
Confidence
89% confidence
Finding

The logo-design example may involve proprietary brand concepts, company names, or unreleased identity work being sent to an external AI provider. In this skill context, the danger is elevated because branding and creative assets are often confidential, so normalizing direct submission to a third party without explicit consent or data classification checks poses a genuine confidentiality risk.

Content

Scanner excerpt · prompt.md (reported line 151)May include surrounding context.

你的执行:

bash
curl -X POST https://api.lovart.ai/v1/design/generate \
  -H "Authorization: Bearer $LOVART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

触发词包含“AI 设计生成”“生成产品图片”“创建营销素材”等较宽泛表述,可能在与该技能无关的普通对话中被误触发。一旦触发,后续流程可能引导用户向外部第三方服务发送提示词、业务素材或图片内容,增加非预期数据外发风险。

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · skill.md (reported line 86)May include surrounding context.

示例 1: 生成产品图片

bash
curl -X POST https://api.lovart.ai/v1/design/generate \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

Natural-language metadata in the description and triggers appears to require a Chinese-language interaction model, but the manifest does not state that the skill is China-specific or that users may choose another language. This can conflict with organizational language-choice policies when no opt-in or justification is provided.

Content

No source excerpt is available for this finding.