Security checks for vulnerabilities and agentic risk
Overview
This skill is a documentation generator that reads project code and writes project documentation; its sensitive behavior is disclosed and aligned with that purpose.
Install this only if you are comfortable with a skill reading your repository to generate or synchronize documentation. Review the generated docs before committing them, especially OBSERVATIONS.md and any inferred content marked 〔INFER〕; avoid opening the SVG in an environment where clickable sendPrompt actions can run unless you want those prompt suggestions.
This duplicate finding still corresponds to the same onclick sendPrompt(...) behavior. The risk is not the displayed text itself, but that a static asset can predefine agent prompts and influence workflow when interacted with, which is a meaningful control-channel issue in an agentic environment.
Content
Scanner excerpt · skill_coverage_map.svg (reported line 40)May include surrounding context.
This duplicate finding still corresponds to the same onclick sendPrompt(...) behavior. The risk is not the displayed text itself, but that a static asset can predefine agent prompts and influence workflow when interacted with, which is a meaningful control-channel issue in an agentic environment.
Content
Scanner excerpt · skill_coverage_map.svg (reported line 40)May include surrounding context.
This duplicate likewise reflects interactive prompt injection through sendPrompt(...). In the context of a documentation-generation skill that may inspect source code and UI artifacts, such embedded prompts could bias analysis, trigger unintended data processing, or bypass explicit user framing.
Content
Scanner excerpt · skill_coverage_map.svg (reported line 51)May include surrounding context.
This duplicate likewise reflects interactive prompt injection through sendPrompt(...). In the context of a documentation-generation skill that may inspect source code and UI artifacts, such embedded prompts could bias analysis, trigger unintended data processing, or bypass explicit user framing.
Content
Scanner excerpt · skill_coverage_map.svg (reported line 51)May include surrounding context.
The trigger list is extremely broad and includes common phrases like '生成文档', '更新手册', and '同步文档', which can match ordinary user requests across many contexts. This increases the chance of unintended skill invocation, causing the agent to apply this skill's workflow and constraints when the user did not explicitly want reverse-engineering or documentation generation behavior.
Content
No source excerpt is available for this finding.
Natural-Language Policy Violations
Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding
The manifest description is entirely in Chinese and line L188 further constrains filenames to use the same language as the user dialogue, while the skill does not state that users may choose their preferred language. Under the policy, forcing a specific language or locale without opt-in is a natural-language policy violation unless clearly justified as region-specific.
Content
No source excerpt is available for this finding.
Natural-Language Policy Violations
Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding
The document is entirely written as mandatory instructions in Chinese, beginning with the title and directive to extract information across seven dimensions. Under the policy, forcing a specific language without offering a language or locale choice is a natural-language policy violation.
Content
No source excerpt is available for this finding.
Natural-Language Policy Violations
Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding
The file explicitly requires all diagram text to be in Chinese, which is a language policy constraint stated in natural language. It does not offer an alternative language choice or indicate that the document is limited to a justified Chinese-only regional context.
The title and guidance are written entirely in Chinese, and later instructions explicitly prioritize extracting Chinese strings and i18n Chinese files as the primary source of UI text. This imposes a specific language/locale assumption without any user opt-in or explanation that the skill is intended only for Chinese-language projects.
Content
No source excerpt is available for this finding.
Session Persistence
Medium
Category
Rogue Agent
Confidence
75% confidence
Finding
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
Content
Scanner excerpt · references/exploration-strategy.md (reported line 86)May include surrounding context.
The file title and all operational instructions are written in Chinese, and there is no statement offering multilingual output or user language preference. Because this guidance is framed as a general-purpose process for '任何项目类型，任何语言' rather than a China-specific workflow, the implicit language requirement is a natural-language policy concern.
Content
No source excerpt is available for this finding.
Natural-Language Policy Violations
Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding
The SVG’s visible labels and interaction text are predominantly in Chinese, including the title, section labels, and bottom note. Under the policy rule for language/locale, this is a natural-language constraint that does not offer opt-in or alternative locale handling, and the file does not document a region-specific justification.
