Security audit
Security checks for vulnerabilities and agentic risk
The skill is documentation-only, but its USA branding conflicts with Estonia-focused guidance and its broad implicit triggers could route users into the wrong business/legal context.
Review before installing. This skill does not appear to execute code or access sensitive data, but its name and routing metadata are misleading for a business/legal planning context. Install only if you specifically want Estonia e-Residency and Estonian OÜ workflow guidance, and consider disabling implicit invocation or renaming/rescoping it first.
The activation description is broad enough to match vague requests for 'business-and-operations,' 'analysis,' 'workflow,' or 'implementation support,' even when they are unrelated to this niche topic. Overbroad triggering can make the agent invoke this skill in the wrong contexts, leading to irrelevant or incorrect advice and increasing prompt-surface exposure from unnecessary skill loading.
The default prompt is overly broad and vague, combining a generic invocation pattern with truncated task wording. This can cause the skill to activate in contexts the user did not clearly intend, increasing the chance of misrouting user requests or injecting the skill into unrelated conversations.
The trigger phrases are very generic and partially malformed, using broad language like 'Help me' and 'I need a practical workflow' tied to a long, loosely scoped requirement string. This increases the chance the skill is invoked for unrelated business, migration, or Estonia questions, which can cause prompt-routing mistakes and unintended execution of the skill in contexts the user did not explicitly request.
The trigger phrases are broad and loosely scoped, including generic terms like business-and-operations and practical workflow, which can cause the skill to activate in contexts unrelated to Estonia e-Residency or business migration planning. In an agent ecosystem, unintended invocation can route user queries to the wrong workflow, causing inappropriate guidance, context leakage into an irrelevant skill, or unsafe overreach in automated task selection.
The skill’s identity and advertised purpose say 'USA Business Migration Planner,' but the actual requirement and workflow are about Estonia e-Residency and operating an Estonian OÜ. This mismatch can cause the wrong skill to activate, mislead users about scope, and route sensitive business/legal questions into an unrelated workflow, increasing the chance of bad guidance or unintended data disclosure.
The keyword list includes generic words like 'running' and 'surprised,' which commonly appear in ordinary conversation and are not unique to this skill’s domain. This raises the risk of accidental activation, causing context confusion, degraded answer quality, and possible disclosure of user inputs to an irrelevant skill path.
The sample trigger phrases are formulated so broadly that they can normalize activation on partial, everyday language rather than clear domain-specific intent. In practice, this can train downstream routing or users to invoke the skill for loosely related requests, amplifying misfires and incorrect assistance.
该技能文件为 SKILL.zh-CN.md，全文以简体中文面向用户给出说明，但未看到提供语言选项、用户偏好选择或对中文限定适用性的自然语言说明。根据规则，若技能在语言/地区上形成默认强制而无用户选择，可能构成语言/locale 策略问题。
The skill metadata and title claim a USA business migration planner, but the body actually guides users on Estonia e-Residency and operating an Estonian OÜ. This scope mismatch can cause the wrong skill to be invoked or trusted under false pretenses, leading to misrouting, incorrect advice, and reduced operator visibility into what the skill really does.
The title contradicts the substantive scope described in the document, which increases the chance that users or orchestrators will rely on an incorrect label when selecting the skill. In an agent setting, misleading scope is dangerous because it can trigger unintended workflows and produce jurisdictionally wrong operational guidance.
The trigger keywords include broad terms like business-and-operations, surprised, running, and residency, which can overlap with ordinary conversation and unrelated tasks. Overbroad triggers can cause accidental activation, exposing users to irrelevant or incorrect specialized guidance and increasing the blast radius of the skill's existing scope confusion.
The sample invocations use generic phrasing that lacks strong domain constraints, reinforcing weak activation boundaries. In practice, this can train users or calling systems to invoke the skill on loosely related prompts, resulting in mistaken execution and unreliable advice.
The skill branding and prompt anchor the interaction to a specific locale ('USA') while the underlying description references Estonia e-Residency and Estonian OÜ operations. This mismatch can steer users into jurisdiction-specific advice without explicit opt-in or clarification, creating a risk of incorrect or inapplicable guidance.
Implicit invocation is enabled even though the skill lacks clear activation constraints or narrowly scoped trigger language. This makes accidental or non-consensual activation more likely, which can lead to inappropriate handling of user queries and unexpected tool selection.
The trigger sentences are broad, unnatural, and likely to match loosely related user requests, which can cause the wrong skill to activate outside its intended scope. In an agent system, this increases the chance of prompt-routing errors, irrelevant automation, and accidental handling of unrelated business or migration topics under the wrong workflow.
This file is presented as a Chinese README, but key user-facing content such as the demand description, workflow description, and trigger phrases remains in English. That can amount to a language/locale policy issue because the skill experience appears to force or assume English without explicit user opt-in or explanation.
The usage section defines trigger sentences only in English and does not indicate that other languages are supported or that English is merely an example. This can amount to an implicit language constraint without opt-in or justification.
No suspicious patterns detected.
