Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed Datu API poster/image-editing integration that sends user-provided content to an external service with stated consent and sensitive-data checks.
Install only if you are comfortable sending prompts, documents, image URLs, and images to Datu. Use non-sensitive material first, do not upload credentials or confidential documents without approval, and understand that the host may store your Datu API key for reuse and generated results may be retained for about 30 days.
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
# Digilifeform AI 大图生成器
## 触发词
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
# Digilifeform AI 大图生成器
## 触发词
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
# Digilifeform AI 大图生成器
## 触发词
该变更日志说明把“改图、换字、继续修图等请求”作为必须命中 AI 修图接口的触发条件，其中“换字”“继续修图”等表述较短且缺少边界说明，容易与普通对话或泛化编辑请求重叠。文中也提到扩充触发词以提升命中率，但没有给出明确约束、排除条件或负例，存在误触发风险。
The instruction to pass user-provided raw information directly to an external API without summarization or minimization increases the chance that sensitive data, regulated content, or embedded credentials are transmitted verbatim. In this skill's context, the destination is a third-party service, so preserving all raw input can defeat data-minimization safeguards and magnify privacy and compliance risk.
The file upload example explicitly says the system will parse uploaded files, but it does not warn users that document contents will be transmitted to and processed by the service. This creates a real privacy and data-handling risk because users may upload sensitive internal or personal documents without understanding that their contents are being extracted and used as prompt material.
The examples use external image URLs for image editing and reference inputs without stating that the backend will fetch those URLs and process the image contents. This omission can expose private or internal resources, surprise users about third-party retrieval of content, and encourage unsafe use of URLs that may reveal sensitive assets or trigger unintended network access by the service.
The skill documentation is written as a Chinese-only skill title and all operating instructions are presented in Chinese, with no indication that users may choose another language or locale. Under the stated policy, forcing a specific language without opt-in is a natural-language policy violation unless the locale restriction is explicitly documented and justified.
The trigger list contains broad everyday phrases like '修改图片', '换字', and '替换 logo' that can match many ordinary image-related requests and cause this skill to be invoked when the user did not specifically ask for this external service. Because the skill is designed to send text, files, or image URLs to a third-party endpoint, over-broad routing increases the chance of unintended data transfer or credentialed API use without sufficiently specific user intent.
The description explicitly states that user text, documents, and existing posters/images are sent to the Datu external API, but the manifest does not present a prominent, user-facing warning at the point of invocation about third-party data transfer and retention. Because this skill handles potentially sensitive documents and images and retains generated results for 30 days, users may unknowingly disclose confidential material to an external processor.
The manifest registers very broad trigger phrases such as generic image-editing requests ('改图', '修改图片', '换字', '调整颜色') that can match ordinary user intent without clearly signaling that files and prompts will be sent to an external third-party API. This creates an overbroad invocation surface that can cause unintentional routing of sensitive images or documents to the external service, especially in hosts that auto-suggest or auto-invoke skills based on trigger words.
文件标题及全文说明均固定为中文，未见提供语言/locale 选择或说明该语言限制的适用范围。按规则，强制特定语言而无用户选择可能构成自然语言层面的语言/区域策略问题。
All instructional text and usage examples are presented in Chinese, with no indication that this language constraint is optional or region-specific. Under the language/locale policy, forcing a single language without user opt-in can constitute a natural-language policy issue.
This SVG contains multiple visible text strings exclusively in Chinese, including the title, subtitle, features, and statistics labels. Because SQP-3 applies to all file types and flags language-policy issues, this is a natural-language locale constraint with no indication of user opt-in or justification for a Chinese-only presentation.
No suspicious patterns detected.