Security audit
Security checks for vulnerabilities and agentic risk
This is a local document scanner that reads user-selected files, reports masked PII/secret matches, and only writes a cleaned copy when explicitly requested.
Installers should understand that this is a local regex-based aid, not a full DLP system. Use it on files you intend to scan, review masked findings before sharing documents, and be careful with --clean --out because it writes a normalized copy and does not remove secrets from the source text.
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
# Clean Note
This file has no secrets or PII.
Just a plain paragraph.
The natural-language description specifies scanning for CN-mobile and CN-ID data types, which imposes a locale-specific policy choice in the skill behavior. The README does not present this as optional or explain that the skill is intended only for a China-specific compliance context.
Suspicious Unicode normalization or mixed-script content
This code writes normalized document content to a user-specified path when --clean is enabled, but the operation has no confirmation prompt, warning print, or explanatory comment/docstring near the write. For a code-file warning review, this is a safety-relevant file write that proceeds silently until after completion.
The changelog states "EN-first, ZH after", which is a natural-language policy choice favoring a specific language ordering. There is no indication that users can choose their preferred language presentation or that this locale preference is required by the skill's function.
The CLI description includes Chinese text alongside English, and the skill metadata/docstring emphasizes Chinese-specific compliance patterns without stating that the locale is optional or region-specific by design. Under the policy, locale-specific behavior should be opt-in or clearly justified when constrained.
No suspicious patterns detected.