Use this skill whenever the user needs a pure md/txt distributable version of the WC26 Predict project that can reconstruct a runnable local prediction app without downloading any upstream repository. This skill is specifically for rebuilding the text-only ELO-capable WC26 project from bundled markdown references. Also use it when the user says "只支持 md/txt", "给朋友发一个纯文本 skill", "安装后重建 WC26 预测项目", "不要依赖上游仓库", or wants the earlier modified project packaged in a text-only skill format.
Use when designing or improving AI engineering workflows after the stack direction is already mostly known. Covers prompt pipelines, MCP integrations, tool-using agents, reusable workflow specs, evaluation loops, and workflow decomposition. Trigger this for agent architecture, prompt refinement, tool grounding, workflow design, and turning repeatable AI tasks into durable systems. If the main question is local model selection, deployment path, or LM Studio versus Ollama versus MLX, use local-ai-systems-studio instead. If the main request is to create, rewrite, benchmark, or improve a skill itself, use skill-creator instead even when the skill is AI-related.
Distill a bestie into an AI Skill. Auto-collect Feishu/DingTalk data, generate Work Skill + Persona, with continuous evolution. | 把闺蜜蒸馏成 AI Skill，自动采集飞书/钉钉数据，生成 Work + Persona，支持持续进化。
Distill a digital self into an AI Skill. Auto-collect Feishu/DingTalk data, generate Work Skill + Persona, with continuous evolution. | 把分身蒸馏成 AI Skill，自动采集飞书/钉钉数据，生成 Work + Persona，支持持续进化。
Use when packaging GitHub projects for jobs, internships, interviews, or public portfolio use. Trigger this for README rewrites, project positioning, resume bullets, interview story extraction, repo cleanup, feature highlight framing, and turning raw coursework or side projects into stronger showcase material.
Use when working on Java full-stack engineering tasks, especially Spring Boot, MyBatis or JPA, SQL, Vue or React admin systems, backend APIs, performance issues, refactors, or interview-grade project hardening. Trigger this for feature development, bug fixes, code review, schema tuning, API cleanup, and turning student or side projects into stronger engineering work.
Use when turning a local AI idea, issue, or workflow request into a concrete system plan and the main uncertainty is still about the local stack itself. Covers task scoping, local model and tool evaluation, deployment choices such as MLX, GGUF, Ollama, LM Studio, or vLLM, hardware fit, workflow direction, and issue-to-plan conversion. Trigger this for local LLM selection, tool comparison, deployment tradeoffs, setup planning, and "how should I build this locally" requests. If the stack is already known and the main work is deeper workflow orchestration, use ai-agent-workflow instead.
Use when creating or improving Word, WPS, PPT, PDF, report, proposal, summary, or presentation materials. Trigger this for document cleanup, report drafting, slide deck structuring, PDF conversion, meeting summaries, project presentations, and polished office-style deliverables.