tense-i

Use when analyzing an existing repository to reconstruct missing or outdated agent-facing prompt specs from code, tests, docs, and configuration, especially before planning, onboarding, repo handoff, or coding-agent execution.

提供 Cocos Creator 2.4 游戏引擎的全面开发指导，包括组件系统（cc.Class、cc._decorator）、生命周期回调、事件系统、Asset Manager 资源管理、缓动/动作系统、对象池、UI 系统、物理碰撞以及可试玩广告优化。在用户编写或重构 Cocos Creator 2.x TypeScript/JavaScript 代码、实现游戏功能、处理资源加载与释放、优化性能/包体大小、审查代码变更、搭建可试玩广告项目架构时触发。也适用于用户提到 cc.Class、cc.Component、cc.Node、cc.resources、cc.assetManager、cc.tween 等 2.x API 时使用。

提供 Cocos Creator 3.8 游戏引擎的全面开发指导，包括组件系统（_decorator、Component）、生命周期回调、事件系统（EventTarget、input）、resources 资源管理、tween 缓动系统、对象池、UI 系统、物理碰撞以及可试玩广告优化。在用户编写或重构 Cocos Creator 3.x TypeScript 代码、实现游戏功能、处理资源加载与释放、优化性能/包体大小、审查代码变更、搭建可试玩广告项目架构时触发。也适用于用户提到 import from 'cc'、Component、Node、resources、tween、director 等 3.x API 时使用。

Cocos 프로젝트에 MCP 서버 설치

Sync planning and execution outcomes from superpowers into Plane, and create an end-to-end closed loop through status updates: Requirements/Planning -> Plane Work Items -> Implement -> Verify -> Accept -> Close. Supports the full project lifecycle: initialization, new features, bug fixes, technical debt, hotfixes, and related scenarios.

Validates architecture documentation completeness by checking for technology stack, API specifications, database schema, security architecture, and alignment with feature specification. Ensures plan.md is complete before implementation.

Guide for using code execution capabilities to perform parallel batch processing, conditional filtering, and data aggregation. This skill should be used when agents need to analyze multiple files efficiently, validate large result sets, aggregate data from multiple sources, or reduce token consumption through execution-based filtering. Provides reusable templates for quota-aware workflows, error handling patterns, and token-efficient data processing.

Execute codeagent-wrapper for multi-backend AI code tasks. Supports Codex, Claude, and Gemini backends with file references (@syntax) and structured output.

generate production-grade interview question banks from a role, jd, technology direction, project, or business system. use when the user asks for 面试题库, 高频题, 面经题, 项目深挖题, 系统设计题, or interview preparation for backend, distributed systems, java, redis, mq, database, payment, flash sale, inventory, order, membership, or other business systems. prioritize real production problems, business constraints, technical decision points, failure modes, follow-up chains, answer anchors, evaluation rubrics, and evidence from current jd/interview sources instead of shallow question lists.

analyze current job market demand for a user-specified career direction using real job descriptions from the last 30 days. use when the user asks to evaluate a role, technology direction, career path, market demand, salary value, job requirements, or skill gap based on boss直聘, lagou, liepin, linkedin, indeed, company career pages, or user-provided jd text. must collect and analyze at least 6 valid recent jds before making strong conclusions; every conclusion must be grounded in jd evidence.

orchestrate a personal ai-era growth operating system for daily judgment training, weekly theme planning, information intake, scenario analysis, language output, market calibration, and social practice. use when the user asks to run, design, review, or maintain a daily mustdo system; build a personal ability operating system; plan weekly/monthly learning themes; coordinate hacker news, github trending, track awesome list, scenario harness, boss/job-market review, or workplace/social-practice reflection; or convert scattered self-improvement tasks into a closed-loop training workflow.

analyze technical decision points in a business scenario or system design. use when the user asks to identify, compare, review, or document key technical decisions for systems such as flash sale, payment, order fulfillment, inventory, messaging, recommendations, workflows, risk control, or other business systems. especially useful for turning a scenario into decision points, trade-off matrices, failure modes, recommended choices, architecture constraints, and coding-agent-ready implementation guidance.

以“解决方案架构师 + 人类判断力教练 + harness 工程师”的复合视角，把秒杀、IM、清结算、订单履约、推荐、内容社区等业务场景拆成结构化架构分析、人类判断文档、难点深拆和可执行 prompt 包。Use when: 用户要求系统设计、业务场景拆解、架构方案、harness prompt、让 Claude Code/Codex/Cursor 实现，或希望用 AI 作为工程杠杆但不外包判断力。输出强调人类必须亲自拥有的问题定义、不变量、取舍、失败模式、验收标准和复盘问题，再把确认后的判断压成 agent 可执行契约。

将一段口语化、碎片化的每日热点串讲整理成结构化中文分析文档，重点识别热点叙事与关键事实之间的信息差、背景缺口、误读点和后续观察指标。 Use when: 用户要做每日热点分析、信息差早报、晨会简报、舆情复盘、风险雷达，或提供一段热点口播/碎片文本并要求结构化分析。

输入一个项目或子模块，输出基于DDD的“系统业务架构与技术规划分析文档”。文档必须足够详尽、不可过度精简；最终总行数必须 >1000 行；内容过长时必须按“分片追加协议”多次输出并保持行号连续。

对 GitHub Trending 仓库或单个 GitHub repo 输出逐仓库同等深度的中文分析，覆盖详细 TL;DR、核心特性、痛点到方案到效果、证据来源和快速试用路径。 Use when: 用户提到 GitHub Trending、热门仓库、想快速读懂某个 repo、想知道它解决什么问题、为什么火、有哪些核心特性，或要求做榜单速报。

Automatically discover protocol skills when working with HTTP, TCP, UDP, QUIC, and network protocols

Analyze application and system logs to identify errors, patterns, and root causes. Use log aggregation tools and structured logging for effective debugging.

Implement structured logging with JSON formats, log levels (DEBUG, INFO, WARN, ERROR), contextual logging, PII handling, and centralized logging. Use for logging, observability, log levels, structured logs, or debugging.

Design property-based tests that verify code properties hold for all inputs using automatic test case generation. Use for property-based, QuickCheck, hypothesis testing, generative testing, and invariant verification.

将 Spec-Driven Development（Spec Kit / SDD）+ Vibe Coding 的 spec/plan/tasks 产物自动落到 Plane， 并通过状态回写形成端到端闭环：PRD -> Spec -> Plane Work Items -> Implement -> Verify -> Close。 支持项目全生命周期：初始化、新需求、缺陷修复、技术债务、紧急修复等场景。