Ae1
High
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md 主 `SKILL.md` 只放通用规则和路由。产物细则必须按需读取 `references/`，不要一次性加载所有 ref。
Security audit
豆包创意设计
Security checks for vulnerabilities and agentic risk
Overview
This is a markdown-only image design helper whose image-generation and image-editing behavior matches its stated purpose, with some language and routing caveats.
Install only if you want a broad Doubao-oriented image design prompt router. For English, multilingual, legal, educational, or exact-brand-text work, explicitly tell the agent to preserve the requested language and exact visible text because the skill defaults strongly toward Chinese prompt conventions.
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
- Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (16)
Ae1
High
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md 主 `SKILL.md` 只放通用规则和路由。产物细则必须按需读取 `references/`，不要一次性加载所有 ref。
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The skill description is extremely broad and overlaps with many generic image-related requests, making it likely to preempt more specialized or safer skills. In an agentic environment, overly broad routing can cause incorrect tool selection, unintended access to image-generation/editing capabilities, and policy bypass through misclassification of user intent.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 98% confidence
- Finding
Forcing Chinese-only prompt generation without user opt-in can override user intent and create unsafe prompt translation or policy interpretation errors, especially when source instructions, legal text, or safety-critical labels are in another language. This is mainly an integrity and usability risk rather than a direct security exploit, but it can degrade output fidelity and cause the agent to mishandle multilingual content.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 99% confidence
- Finding
The prompt rules hard-code that final prompts must be continuous Chinese text, which imposes a fixed language policy across all tasks. In contexts involving multilingual branding, exact text rendering, compliance copy, or user-supplied non-Chinese content, this can distort instructions and introduce mistakes in generated outputs.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 94% confidence
- Finding
The file repeatedly mandates that all visible text be wrapped in Chinese quotation marks and, in examples and formatting rules, consistently requires Chinese-language textual conventions. Because this is a natural-language instruction in a markdown file and no opt-in or region-specific justification is provided, it creates a locale/language policy concern under the rule for forced language constraints.
- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
L003 的 description 使用了很长的自然语言枚举，并以“用户当前请求或近 1 轮对话涉及…时触发”作为激活条件，范围包含“角色设计”“剧情角色”“主题套图”等宽泛表达，容易与一般设计对话重叠。该描述没有给出负例、上下文限制或更明确的调用边界，因此存在误触发风险。
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 91% confidence
- Finding
L251-L275 多处规定所有渲染文字“必须”使用中文双引号包裹，并以中文规范统一约束文本输出，但未说明这是用户可选项，也未给出仅适用于特定中文品牌/地区场景的限制。该自然语言要求可能在非中文用户或需其他语言排版的任务中强制施加特定语言/locale 规范，构成语言策略问题。
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The instruction "所有画面文字必须加中文双引号" imposes a specific language/locale formatting requirement on all outputs. This is a natural-language policy concern because it forces Chinese-style quotation marks regardless of the user's preferred language or the brand's target locale, and no opt-in or exception is provided.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
This line requires every piece of visible text to be wrapped in Chinese quotation marks, creating a blanket locale constraint across all generated content. Because the file does not offer a user choice or justify a China-specific scope, this constitutes a language/locale policy violation.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 84% confidence
- Finding
The instruction requires all on-image text guidance to use Chinese quotation marks “”, and later sections repeat this as a mandatory rule. Because this file also contemplates English and other foreign-language deliverables, enforcing Chinese-specific punctuation is a locale-specific constraint not presented as optional user choice.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 90% confidence
- Finding
This line states that all text to be rendered must be enclosed in Chinese quotation marks “”. That is a hardcoded locale-specific formatting requirement and can conflict with user-requested English or other non-Chinese outputs without any opt-in.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 90% confidence
- Finding
The file repeats a strict requirement that all explicit text be written using Chinese quotation marks. Since the skill supports multilingual e-commerce materials, this is an unjustified forced locale convention rather than a user-selected option.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The skill hard-codes Chinese output conventions for all rendered text and prompt construction without preserving user language preference or offering opt-in. This can override user intent, cause incorrect language transformation of user-provided content, and create prompt-safety issues where exact text fidelity matters, especially for educational diagrams, labels, or multilingual inputs.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The prompt rewriting rules require the final generation prompt to be in Chinese and mandate Chinese quotation formatting for all on-image text, regardless of the user's language or the source material. In a knowledge-visualization skill, this can silently alter educational content, formulas, labels, or terminology, reducing accuracy and making the output misaligned with user requirements or downstream tooling expectations.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 94% confidence
- Finding
The instruction mandates that all rendered visible text must use Chinese quotation marks, which imposes a specific language/locale formatting rule regardless of user preference. The file does not offer an opt-in choice or limit this requirement to clearly justified Chinese-only contexts.
- Content
Static analysis
No suspicious patterns detected.