Security audit
Security checks for vulnerabilities and agentic risk
This skill is a small local disclosure-text generator with documentation quality issues but no evidence of hidden access, persistence, exfiltration, or destructive behavior.
Safe to consider for local disclosure drafting, but review the generated text yourself and do not rely on it as legal advice. The publisher should align the docs with the actual CLI and clarify whether file modification, metadata injection, WebFetch, and bilingual output are optional or required.
The file title and all operating instructions are written in Chinese, and there is no indication that the user or agent may choose another language for interaction. Under the stated policy, forcing a specific language without opt-in is a natural-language policy concern unless the locale constraint is explicitly justified.
The top-level description presents the skill as generating compliant disclosure statements and metadata labeling guidance. Elsewhere in the file, the described flow includes auto-detecting AI involvement, writing metadata into content headers/footers, platform-specific formatting, and exporting a publish-ready file, which is materially broader than a disclosure-text generator. This creates a semantic mismatch in stated purpose versus described behavior.
The documentation instructs users to run 'python ai_content_discloser.py --mode auto --language cn_en --input_file content.txt' with parameters like '--input' and '--include_metadata'. But the manifest declares the entry point as 'scripts/ai_content_discloser.py' and the input schema contains only type, ai-level, brand, tool, platform, date, and json. This is not mere incompleteness; the documented invocation model conflicts with the declared interface.
The usage text hard-codes --language cn_en, and the surrounding description presents bilingual zh+en generation as the default behavior rather than an explicit user-selected option. This can violate language/locale policy expectations because the skill appears to impose a specific language combination without opt-in or a clearly justified locale restriction.
The flow documentation states the tool will 'Auto-detect AI involvement level,' 'Write implicit metadata ... into header/footer,' and 'export the publish-ready file.' However, the manifest's actual CLI schema only defines structured inputs like type, ai-level, brand, tool, platform, date, and a stdout text output, with filesystem/network/subprocess permissions all set to false. This is an active contradiction between the documented behavior and the declared implementation surface.
The flow says the tool will 'generate explicit zh + en statements,' which reads as a mandatory language behavior rather than a configurable option. Because the file does not clearly frame this as user choice or narrowly justified scope at that point, it fits the language/locale policy violation category.
The Chinese instructions say the system automatically generates corresponding Chinese and English disclosure text, which imposes a fixed locale pattern in natural-language guidance. The file does not clearly present this as optional at that point, so it may conflict with organizational language-choice policy.
The file states in Chinese that this is the AI-readable skill index and that agents should read SKILL.md first, indicating a language-specific instruction. Because the file does not offer an alternative language or clearly justify that the skill is restricted to Chinese-speaking users, this may violate the language/locale policy for natural-language content.
This code always generates and prints an English disclosure statement in addition to Chinese, regardless of the user's language preference or locale. The policy requires flagging language or locale constraints when a specific language is forced without explicit opt-in or a documented justification.
