Security audit
Security checks for vulnerabilities and agentic risk
This is a Chinese-language expert-advice skill made of Markdown reference files, with broad activation wording but no code, persistence, credential access, or hidden execution behavior.
Install only if you want a Chinese-language business and strategy analysis framework. Use extra judgment for legal, financial, policy, and viral-marketing recommendations, and verify current facts and regulated advice with qualified sources before acting.
The skill description and trigger keywords are extremely broad, covering generic terms like '专家', '风险评估', '营销策略', and '决策支持' that could match a wide range of normal user requests. This can cause the skill to activate outside its intended scope, leading to inappropriate interception of conversations, response hijacking, or unintended instruction precedence over more suitable skills.
The skill content is written entirely in Chinese and prescribes output behavior without offering language negotiation or stating that it is limited to Chinese-speaking users. In multilingual environments this can override user expectations, degrade usability, and create routing or comprehension failures, especially if the skill is invoked for non-Chinese requests.
The trigger phrases for "科技必然推演" include broad conditions such as asking about a technology or business's future trend, entering a new sector, or choosing a technical route. These are common advisory scenarios and the file does not provide exclusions, narrow invocation phrases, or negative examples to clarify when this skill should not activate.
The activation conditions for "创客生态链接" cover generic requests such as learning about a technical community, finding partners/resources, or assessing ecosystem activity. Without a defined trigger phrase set or constraints, these conditions are broad enough to collide with ordinary collaboration and research queries.
This markdown file defines activation conditions such as '用户询问某个行业/领域的政策环境' and '用户面临政策风险的评估需求', which are very broad and lack limiting context. The description does not provide explicit exclusions or negative examples, so the skill could be invoked for ordinary discussion about policy or industry topics.
Phrases like '用户需要评估经济周期位置' and '用户希望制定穿越周期的战略' can overlap with common business-advice conversations and do not clearly distinguish when this skill should activate. The file also omits exclusion conditions or examples of requests that should not trigger the skill.
The trigger conditions for '商业模式升维' are broad and overlap heavily with common business-advice requests, so the skill may be invoked in situations where its assumptions or scope do not fit. This can cause unintended routing, over-application of a single expert persona, and decision support that appears authoritative without clear boundaries or handoff criteria.
The trigger conditions for '商业避坑防爆' are similarly vague and can match many generic decision-support situations, which increases the chance of accidental invocation. In a multi-skill agent, ambiguous routing can distort outputs, create inconsistent analysis paths, and let a broad advisory skill handle requests that should instead go to domain-specific, legal, compliance, or safety-focused workflows.
The trigger conditions for Skill 1 are written as broad user situations such as '需要评估或优化供应链风险' and '担心供应链中断影响业务' without explicit boundaries, invocation phrases, or exclusion criteria. This can overlap with many ordinary business discussions and does not clearly specify when the skill should activate versus when it should not.
The trigger conditions for Skill 2 include general statements like '需要分析竞争格局' and '需要寻找差异化定位,' which are high-level business needs rather than precise activation conditions. Without concrete trigger phrases, contextual limits, or negative examples, the skill could be invoked unintentionally during broad strategic conversations.
The trigger conditions are broad enough that ordinary branding or marketing queries could activate this skill unexpectedly, causing the agent to steer users toward aggressive conflict-driven advertising tactics without clear user intent. In this file, that matters more because the skill explicitly promotes controversy, emotional provocation, and attention capture, which can amplify harmful or manipulative outputs when invoked unintentionally.
The social virality trigger is vague and overlaps with routine requests for content planning or campaign design, so the skill may activate for normal marketing assistance and push users into viral-growth tactics they did not specifically request. This context increases risk because the skill advocates engineered emotional tension, social-chain exploitation, and multi-version testing to maximize spread, which can facilitate manipulative engagement strategies at scale.
The trigger conditions for Skill 1 are broad and generic, covering common requests like planning AI architecture or assessing deployment challenges. In an agent environment, this can cause unintended invocation during ordinary technical discussion, leading the system to apply a specialized decision framework when the user did not explicitly request it, which may skew outputs or broaden data handling beyond user expectations.
The trigger conditions for Skill 2 are similarly vague and overlap with ordinary maintenance, architecture, and technical risk discussions. This creates a prompt-scope weakness where the skill may activate outside its intended context, potentially biasing recommendations, causing unnecessary workflow branching, or surfacing specialized guidance when a simpler response was expected.
Skill 1 的触发条件如“解决一个技术或设计问题”“产品开发中遇到瓶颈”范围非常宽，容易在大量普通请求中被误触发。对一个会输出强方法论和决策框架的技能来说，过宽匹配会导致路由偏置、错误介入用户任务，并放大不适用建议被当作权威结论采纳的风险。
Skill 2 的触发条件如“优化用户体验”“平衡产品功能和成本”同样较模糊，覆盖面接近通用产品咨询。由于该技能会对技术、体验和成本进行结构化权衡，若在缺少边界时自动调用，可能对不相关场景产生过度自信的建议，影响后续决策质量，尤其在产品或业务流程中会形成系统性误导。
The trigger conditions for '张三风控模拟' are broad enough that the skill could be invoked for many loosely related requests about business behavior, disputes, or compliance review. In an agent setting, ambiguous activation can cause the model to apply a legal/compliance persona when the user did not clearly request it, leading to overreach, inappropriate legal-sounding guidance, or unintended handling of sensitive risk scenarios.
The second skill's activation conditions are similarly vague, covering generic needs like compliance awareness training or culture building without clear thresholds. This increases the chance of unintended invocation in routine business or HR discussions, which can produce authoritative compliance guidance outside the proper scope and without user consent or contextual safeguards.
The activation conditions for "视觉极简降噪" are phrased generically, such as "用户需要优化视觉设计方案" and "用户需要评估视觉信息的清晰度." These lack explicit scope, exclusions, or concrete trigger phrases, so the skill could be invoked for a wide range of normal design conversations unintentionally.
Phrases like "用户需要提升产品的感知价值" and "用户希望设计超越视觉层面" are conceptually broad and do not clearly distinguish this skill from general product or brand design assistance. The file does not provide negative examples or constraints to prevent overbroad matching.
The skill content and operational instructions are presented only in Chinese, which can constitute a language-policy issue when no user opt-in or locale constraint is stated. There is no indication that the skill is intentionally limited to Chinese-speaking users or a China-specific deployment context.
This markdown file is written entirely in Chinese and presents the skill guidance, expert taxonomy, and invocation patterns only in that language. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation when no alternative language or choice mechanism is offered.
No suspicious patterns detected.
