Security audit
Security checks for vulnerabilities and agentic risk
The skill appears purpose-aligned but overstates its compliance-audit capabilities and promotes external compliance services without enough data-handling disclosure.
Treat this as a preliminary checklist tool, not legal advice or a complete compliance audit. Before installing or using the linked API/web app, verify who operates the service and avoid submitting sensitive product plans, user-data maps, legal documents, or marketing drafts unless its privacy and retention terms are acceptable. Confirm all findings with qualified counsel in each target market.
A second description/behavior mismatch indicates the skill may only retrieve or display regulation summaries while claiming substantive compliance assessment across multiple regulatory domains. In a compliance context, overstating assessment capability is especially dangerous because users may make launch or data-transfer decisions based on inaccurate outputs.
A second description/behavior mismatch indicates the skill may only retrieve or display regulation summaries while claiming substantive compliance assessment across multiple regulatory domains. In a compliance context, overstating assessment capability is especially dangerous because users may make launch or data-transfer decisions based on inaccurate outputs.
The skill advertises executable shell capability via scripts/regulations.sh but does not declare any tool scope or allowed-tools restrictions. Undeclared execution capability increases the risk of unexpected command execution or future expansion into shell actions without user visibility or platform policy enforcement.
The trigger list contains broad phrases such as general compliance and privacy-related requests, increasing the chance the skill is invoked in contexts beyond its actual competence. Overbroad routing can expose users to low-quality or misleading compliance guidance when a more specific or safer skill should have handled the request.
POST /check is described as a marketing-content compliance endpoint, which conflicts with the advertised global regulatory compliance audit purpose. This inconsistency can cause users or downstream agents to send sensitive compliance materials to an endpoint that may be processing a different content class than expected.
The skill promotes a real external API backend and web app for compliance checks without warning users that submitted information may be shared with third-party services. Because compliance assessments often involve sensitive launch plans, data flows, jurisdictions, or potentially regulated personal data, undisclosed external transmission creates significant confidentiality and privacy risk.
The documented scope shifts from global regulatory launch compliance into marketing-content banned-word screening and SEO compliance services, indicating hidden or poorly disclosed secondary processing purposes. Scope drift is risky because users may submit sensitive business, product, or user-data details believing they are for one compliance purpose while the backend or web app is oriented toward another.
The skill metadata promises an API-powered regulations database and real backend, but this file relies entirely on hardcoded local rules. In a compliance-checking skill, this mismatch can mislead users into trusting stale or incomplete legal guidance, creating a security/compliance integrity risk through deceptive capability claims.
The manifest advertises compliance gap analysis and a remediation roadmap, but the code only enumerates applicable regulations and rough estimates without assessing existing controls. Users may act on a false sense of completeness, overlooking real gaps because the tool presents itself as more authoritative than it is.
The script emits a note containing Chinese text in parentheses as part of its standard report output. This introduces a specific language requirement into user-facing content without offering a language choice or documenting why the locale is mandatory for all users of the tool.
The script emits Chinese text in its output strings such as "核心法规", "罚款上限", and the comment line also frames behavior in a fixed language context, without offering users a language choice. This creates a natural-language policy issue because the skill implicitly enforces a locale/language rather than allowing opt-in or documenting a justified regional constraint.
The described market coverage is broader than what the script actually supports, omitting claimed regions such as the UK and Australia and collapsing broader regions into a few country codes. In a compliance context, unsupported jurisdictions can lead users to assume checks were performed where none were, causing under-compliance risk.
The script makes outbound HTTP requests with curl to a remote API endpoint, but provides no visible user-facing message, comment, or prompt indicating that it will contact an external service. For a code file, network transmission without any disclosure can surprise users, especially when the destination is not self-evident from the invocation alone.
