ai-skills

Convert rough business requirements into complete, implementation-ready solution blueprints in Chinese. Use this whenever the user asks for 业务实现方案, 技术方案, 架构设计, PRD转实现, 需求拆解, 数据模型, 状态流转, 接口设计, or wants a coding agent to build directly from a high-level requirement, even if the input is incomplete.

Review C#, ASP.NET Core, and .NET 10 pull requests or code diffs with a practical checklist focused on correctness, nullability, async/await, EF Core, DI, security, performance, observability, and tests. Use this skill whenever the user asks for a C#/.NET code review, PR review, diff review, checklist-based review, best-practices validation, 架构审查, or wants actionable findings instead of style-only comments.

Investigate whether a news item, rumor, screenshot claim, social post, statement, image/video description, or link is true, partly true, misleading, false, or still unverified. Use this skill whenever the user asks to 辟谣, 事实核查, 查证真假, verify a claim, analyze a screenshot/link/post, or wants a source-backed judgment instead of a quick opinion, especially for politics, disasters, war, public safety, medicine, finance, law, and other high-risk topics.

根据文字、参考图、公网图片 URL 或本地照片制作可直接打印裁剪的证件照。只要用户提到证件照、报名照、护照照、白底/蓝底/红底换底、1寸/2寸、证件照排版、证件照打印、或想把普通照片变成证件照，都应使用这个 skill。默认优先处理用户已有照片；只有缺图或用户明确要求时，才调用 Liblib 的文生图或图生图。

Classify complex user requests into the right execution loop, define stop conditions, validation checks, token/cost boundaries, and whether to execute or only plan based on current permissions. Use for multi-step coding tasks, automation, monitoring, scheduled work, quality gates, repeated iteration, vague complex goals, or any request where Codex should decide between turn-based, goal-based, time-based, or proactive loops. Do not use for simple factual questions or tiny one-step edits.

Export Markdown files to polished, print-ready PDFs and set up reusable local export scripts. Use this skill whenever the user wants Markdown converted to PDF, asks for one-click export automation, needs reusable styling presets, or wants to tune margins, fonts, images, or page layout for resumes, reports, manuals, proposals, or summaries. Default to a script-based workflow instead of a one-off manual conversion.

从本地照片文件夹中按专业摄影标准筛选出值得精修的照片。先生成联络表，再做多轮挑片，最后把终选片复制到“精修”文件夹。macOS 默认走原生 JXA，Linux 和 Windows 可走 Python Pillow 后端。

Generate images with DashScope Qwen-Image 2.0 from rough user requests and save them locally. Use this whenever the user asks to generate an image, illustration, poster, flashcard, teaching card, cover art, scene image, mascot, product concept, or "文生图"/"画一张图", even if the request is vague. Supports first-time setup via EXTEND.md, fills missing style and size from saved preferences, and asks only for still-missing image parameters.

Build or rewrite professional resumes and CVs from scattered materials, existing resumes, PDFs, job histories, work photos, certificates, or target job postings. Use this skill whenever the user asks to create, improve, quantify, tailor, condense, localize, or polish any resume or CV, even if they only provide rough notes or evidence files. Default to producing truthful, well-structured markdown resumes and a concise version when that would help the user apply faster.

Translate English technical content or a provided URL into idiomatic Chinese. Use when the user asks to translate English text or gives a blog/article link. If a link is provided, download and read it first. Output ONLY the Chinese translation, preserving the original structure and code blocks.