Security audit
Security checks for vulnerabilities and agentic risk
This skill locally turns provided vendor information into a Chinese vendor-risk brief without network access or persistence, though its bundled script contains inactive generic audit helpers.
Install this if you want a local, Chinese-language draft brief for SaaS/API vendor risk review. Run it only on files you intentionally provide, prefer dry-run or stdout for sensitive material, and treat the result as a review aid rather than a formal security or legal approval.
The declared purpose is a vendor-risk briefing skill, but the detected behavior suggests broader filesystem inspection, content scanning, and mode-switching based on spec.json. That mismatch is dangerous because users and orchestrators may invoke the skill expecting a narrow summarization task while it performs more general auditing or scanning actions that touch unrelated files or data.
The script’s primary behavior is a generic local reporting/auditing engine selected by spec mode, including directory, CSV, pattern, and skill audits, which materially exceeds the declared vendor-risk-brief purpose. This capability mismatch is dangerous because it broadens the skill’s operational scope, increases the chance of unintended local file inspection, and can mislead users into trusting outputs as vendor-risk analysis when the tool is actually performing unrelated audits.
The trigger section includes "聚焦实际集成影响" as a standalone activation example, which is broad natural language that could overlap with ordinary discussion rather than a clear invocation phrase. The README also provides only examples and does not state explicit boundaries or negative examples for when the skill should not activate.
The skill advertises executable capabilities via python3 and references local files/templates, but it does not declare an explicit tool scope such as allowed tools or permissions. This creates a policy gap where an agent may grant broader-than-necessary shell/file access, making unintended file reads/writes or script execution more likely in a security-sensitive workflow.
The file presents the skill name, operational instructions, routing guidance, and output structure primarily in Chinese, which effectively imposes a language preference without offering the user any explicit language or locale choice. The policy for this audit flags language-forcing behavior unless the skill provides opt-in or clearly documents a justified locale constraint.
This markdown example output presents all headings and instructions exclusively in Chinese, but does not state that the user selected Chinese or that the skill is region/language-specific. Under the policy, forcing a specific language without user opt-in is a natural-language policy concern.
Multiple user-facing report strings are hard-coded in Chinese, which imposes a specific language on generated output regardless of user preference. The file does not offer a language option or explain that the skill is intentionally limited to a Chinese-language context.
The built-in regex scanning for secrets, private URLs, and dangerous shell patterns introduces a generic code/security scanning capability beyond the declared purpose. While not directly executing anything, it encourages use against arbitrary local files and can expose sensitive snippets in output, making the scope expansion more dangerous in a skill that should focus on vendor integration risk summaries.
The presence of skill package and repository auditing logic allows the tool to inspect local project structure and metadata unrelated to vendor SaaS/API risk assessment. In this skill context, that is risky because it enables broader local enumeration and creates an unexpected analysis surface that users may invoke under the assumption they are only generating a vendor-risk brief.
This markdown file presents all user-facing content in Chinese, and there is no indication that users can opt into another language or that the skill is intentionally limited to a Chinese-speaking audience. Under the language/locale policy, forcing a specific language without user opt-in can be a natural-language policy violation.
No suspicious patterns detected.
