Security audit
Security checks for vulnerabilities and agentic risk
The skill is not deceptive or destructive, but it asks agents to automate public posting, scraping, and customer support without clear approval, privacy, or scope controls.
Install only if you intend to manually supervise its workflows. Require explicit approval before scraping sites, posting or replying from social accounts, or handling customer data, and use approved accounts, channels, retention rules, and human escalation for support responses.
The skill description is very broad and lacks explicit activation boundaries, which can cause the agent to invoke the skill for loosely related requests involving scraping, social posting, or support automation. In a skill with access to browser and web tools, ambiguous routing increases the chance of unintended high-risk actions being performed without clear user intent confirmation.
The skill advertises scraping websites, automating social media actions, and handling support workflows without warning about legal, privacy, platform-policy, or account-integrity risks. Because these are externally impactful actions, omission of warnings and constraints can normalize unsafe use and increase the likelihood of unauthorized data collection, spammy posting, or mishandling of customer information.
The examples suggest invocation phrasing but do not specify formal activation rules, scope limits, or approval requirements. This can lead to overbroad matching where the agent treats similar natural-language requests as authorization to automate sensitive workflows, including web scraping and account actions.
The template describes automated customer replies, intelligent routing, and support across email, web chat, and social media, but it does not warn users about the risks of sending automated responses or routing customer data over external channels. In an enterprise support context, this can lead to privacy, compliance, and reputation issues if sensitive customer information is mishandled or if inaccurate automated responses are sent without review.
The invocation phrase is an open-ended natural-language template that can trigger broad scraping and reporting behavior without clear scope limits, exclusions, or authorization checks. In an automation skill, this increases the chance of unintended use against arbitrary sites, including scraping personal data, violating site restrictions, or producing reports from improperly collected content.
The workflow describes automated scraping, analysis, and file generation but omits any warning about privacy, target-site terms of service, or local handling of collected/exported data. That omission can lead users to collect personal or regulated information and persist it into CSV, PDF, or dashboards without understanding legal, contractual, or security obligations.
The template explicitly instructs automated posting, real-time monitoring, and automatic replies on a company Twitter account without any warning, confirmation step, approval gate, or scope limitation. In an enterprise automation skill, these are impactful external actions that can cause reputational damage, accidental misinformation, spam, or inappropriate responses if triggered or configured incorrectly.
The workflow detection logic only checks for Chinese and English phrases in the user's prompt, which embeds a language constraint into the skill behavior. There is no user opt-in, broader language support, or documented justification for limiting recognition to these locales.
All usage examples are presented as Chinese invocation phrases even though the skill name and surrounding document are otherwise in English, and the file does not state that Chinese is optional or region-specific. This can impose an unexplained language expectation on users without opt-in.
The title and instructions require interaction in Chinese, but the file does not indicate that this is an optional locale, nor does it justify the restriction as region-specific. That can violate language/locale policy when users are not given a choice or informed of the constraint.
All user-facing instructions and labels in this file are in Chinese, and there is no indication that the user can choose another language or that the skill is intentionally restricted to a Chinese-language context. This can conflict with organizational language/locale policies requiring opt-in or explicit justification for locale constraints.
The file presents the skill instructions and workflow only in Chinese and does not indicate that users may choose another language. This can be a natural-language policy concern when a skill appears to force a specific language without opt-in or justification.
