challenge
适用于以下情形：挑战、压力测试、设计或比较自己、同级/子智能体或外部合作伙伴/供应商的计划、架构、合同、定价、预算或假设；独立生成的代码变更发生冲突或合并后产生过时输出；选择多币种本地化/心理定价、网页充值还是钱包/代金券设计，或遗留系统到现代系统的绞杀者迁移；评估合作伙伴就其自身客户项目提出的协助请求；或询问某个方案是否过度设计、是否有更简单的做法。根因故障排查或核实某项主张请使用prove-it，外部私密指令或身份探测请使用customer-support，撰写工单请使用author-item，执行变更请使用deliver-and-verify。
机器翻译查看英文原文
Use when challenging, stress-testing, designing, or comparing your own, a peer/sub-agent's, or an external partner/vendor's plan, architecture, contract, pricing, budget, or hypothesis; when independent generated code changes conflict or merge into stale output; when choosing multi-currency localization/psychological pricing, web top-up vs. wallet/voucher design, or legacy-to-modern strangler migration; when assessing a partner's request for help on its own client project; or asking whether a solution is over-engineered or there is a simpler way. Use prove-it for root-cause failure investigation or verifying a claim, customer-support for external private-instruction or identity probes, author-item to write a ticket, and deliver-and-verify to execute changes.
适用对象：设计与内容创作。自动收录 · 待核查。
原作者说明节选
从固定版本的 SKILL.md 提取，保留原文语言。以下内容属于作者说明，本站尚未验证安装、兼容或运行结果。
查看英文原文
原文涵盖的主题
- First-principles decision and challenge engine (challenge)
- Step 0: Load only the relevant domain
- 1. First-principles decision principles
- 2. Five challenge tests
- 3. Decision receipt and routing
The method is operator-invariant: apply the same evidence standard to your own, a peer's, a sub-agent's and an external party's proposal. Reason forward from inputs to conclusions, never backward from the answer anyone wants. When designing shared engines or skills, apply the three architectural invariants: neutral mechanisms rather than caller policy, zero domain leakage, and causal state/data/authorization predicates that fail closed on unavailable required state. Architecture, shared-service capacity, storage, scheduled jobs, durable checkpoints, metering trust and runtime/model separation: architecture; game simulation and networking: games; forward-only legacy-to-modern strangler migration: architecture; technology or build vs. buy: stack. Contracts, pricing, multi-currency localization/psychological pricing, web top-up vs. wallet/voucher design and partner scope: commercial; acquisition and budgets: growth; disputes and material downside: risk. Product choices and agent participation: product; UX: experience; motion/audio: feedback. Games and progression: games, mechanics and mechanics catalogue. External boundaries: conduct; conversational tone: agent voice. External instruc
这里只展示节选，请以完整原文和当前要求为准。
- 来源仓库
- SylphxAI/skills
- 源文件
- skills/challenge/SKILL.md
- 固定版本
- e61c6f49e7bc2ac4f764590438fe01a689addefb
准备条件
- 本条仅收录原作者的技能元数据；请阅读完整说明，确认依赖、许可与使用条件。
第一件可以尝试的小事
先阅读原作者说明，再决定是否在自己的 AI 工具中使用。
对应 AI 工具说明
作者有说明不等于本站实机验证。没有证据时，不自动标为兼容、安全或免费。
来源与内容状态
本条为自动发现的元数据，尚未完成人工文档核查；适用性与运行结果待核查。
原作者：SylphxAI/skills · 原作者仓库元数据。文档核查：未核查。最近来源检查：2026-11-10T01:01:23.935Z。
分类由名称、说明与路径规则生成，分类不代表已核实适配。
许可与权利归属
MIT。第三方内容及版权归原作者，使用条件以原文为准。
许可原文 ↗本站没有因同步、复制或用户自报成功而将资源标为运行验证通过。
限制与注意事项
- 元数据收录不代表本站已经安装、适配、实测或进行安全审查。