Writing & Editing

Soạn văn bản hành chính Việt Nam đúng thể thức Nghị định 30/2020/NĐ-CP - công văn, tờ trình, thông báo, quyết định, báo cáo, giấy mời, biên bản - gồm Quốc hiệu, tiêu ngữ, số và ký hiệu, địa danh ngày tháng, trích yếu, nơi nhận, chữ ký, phông chữ và căn lề. Use when the user asks to write or format a Vietnamese official/administrative document (công văn, văn bản hành chính, tờ trình, quyết định, thông báo, giấy mời) or to check one against Nghị định 30.

Clarify Chinese, English, or mixed-language explanations, comparisons, plans, summaries, and rewrites through words, diagrams, or interaction. Preserve meaning and the task's language.

Write or review Google Play and App Store listing copy (title, subtitle, short and full description, keywords, release notes) for a Flutter app, within every character limit. Use when the user asks for store description, app title ideas, keywords, ASO, or "write my Play Store listing".

Use when drafting, tailoring, or reviewing a Korean 경력기술서 or career achievement statement for 팀장, 팀 리드, manager, or promotion candidates across business, product/R&D, marketing, sales, customer success, supply chain, production/services, HR, finance, IT/data, facilities/admin, and legal/risk/assurance, including research for unfamiliar roles, explaining why decisions and results matter, MECE responsibility mapping, AI·AX roles, recruiter forms, and overlapping career timelines.

Writes and edits short dramas for vertical and AI video in any language: premises, series bibles, episodes, natural dialogue, shot lists with continuity checks, and script review.

Creates, revises, and reviews technical articles using formal encyclopedic language, numbered Markdown sections, manual tables of contents, coherent taxonomies, practical examples, and source-qualified references. Use for technical or engineering articles and knowledge-base entries, especially Markdown files under content/articles, docs/articles, or articles.

Find the right answer across this repository and the internet — a question, a choice between libraries or approaches, a symptom that may be a known upstream bug, a feature that needs an outside API understood first. Investigates the way /gku:plan does, then reads the documentation, the upstream source, its issues and advisories for the versions this project actually runs. Asks you in the chat, in one batched round, whatever only you can answer, and writes your answers into the result. The...

Read text out of ebooks and documents a section at a time instead of dumping the whole file into context: outline first, then only the chapters needed, then search. One command-line tool (standard-library Python) for EPUB, MOBI, AZW3, FB2, DOCX, ODT, RTF, PPTX, XLSX, HTML, EML, CSV, JSON, Markdown and zip archives of them, plus PDF and legacy DOC, XLS and PPT when poppler, calibre or LibreOffice is installed. Use whenever the user asks to read, quote, summarise, search or extract text from...

Build a task step by step in the current session — from a plan file, a markdown brief, or a sentence. Works through ordered steps, ticking each one off in the task file as it lands, so an interrupted run resumes exactly where it stopped instead of starting over. With --auto it runs the whole cycle unattended — build, review in a fresh context, fix, test, repeat — and opens the pull request at the end, stopping only for something that genuinely needs you. Everything it noticed and left alone...

Use when creating, auditing, optimizing, or rewriting marketing copy: posts, ads, landing pages, emails, webinars, push notifications, product launches. Applies 20 cognitive biases from behavioral economics and evolutionary psychology to engineer persuasion that converts without inventing facts. Self-contained core: a 38-row bias router, 14 power combos, 13 anti-patterns with detection rules, 7 tone styles, cultural adaptation for 12 regions, and a mandatory PROVENANCE/TRUTH GATE. Use ONLY...

Build ATS- and human-optimized résumés/CVs in Typst — exact page control (1-page and 2-page versions), no section ever split across pages, measured page fill, machine-parseable layout, regional and multilingual conventions, adversarial review loop, optional job-search guide. Also builds academic CVs (the exhaustive multi-page genre — publications, funding, teaching, supervision, service) with its own template and gate. Use when the user asks to create, redesign, improve, translate, or tailor...

Rewrites one document — README, AGENTS.md, design note, runbook, spec — so it is free of legacy (no history, no backstory), factual (every claim verified against a primary source or marked, nothing invented), concise (one fact in one place), organized (critical information first), and tuned to a declared reader. Takes --target=agent (default), for an LLM that reads the document as context, or --target=human; the two produce deliberately different documents, each following the measured...

Drafts or updates the living-synthesis "## State of Knowledge" block of ONE mature concept or method note in a scientific wiki, weighing its findings by the evidence pyramid, checking each new or changed entry against the PDF of every work it cites, and presenting the block for human ratification before writing. Triggered by /wiki-synthesize, "synthesize this concept", "state of knowledge for X".

Write product positioning and a messaging framework (a messaging house) from existing product docs, FAB analysis, personas and competitor research. Use when a PMM, founder or PM needs a positioning statement, value proposition, messaging pillars, a tagline or homepage hero options, elevator pitches, or messaging for a new product, a rebrand or a new market.

Write technical text that is short, plain and clear: documentation, READMEs, code comments, commit messages, error messages, interface text and release notes. Also use when the user says "simple english", "de-slop" or "make this readable". Says the one thing that matters, in ordinary words, and leaves the details to the code.

Rewrites text so it reads like a careful, plain-spoken person wrote it, removing AI-writing patterns at the word, sentence and story level while keeping every fact. Cleans hidden characters, scores AI-writing patterns from 0 to 1 before and after, and reviews drafts on request.

Knowledge base and workflow for Java and the JVM in any domain — libraries, backends (Spring Boot, Quarkus, Micronaut), JDBC/JPA, Kafka, CLI, JavaFX, Spark, ML/LLM apps, games (libGDX, LWJGL), Minecraft mods/plugins (Fabric, NeoForge, Paper, mixins), embedded.

Portuguese-first multilingual translation, revision and source-grounded quality assurance. Prioritize Chinese ↔ European Portuguese (pt-PT), including formal news and institutional texts, while supporting Portuguese ↔ other languages and general multilingual pairs when requested.

A method for turning template-shaped deliverables (articles, web pages, decks, docs, READMEs, dashboards, portfolio pieces) into experiences people actually want to consume. USE THIS SKILL whenever the user asks for content to be "fun", "visually engaging", "creative", "out of the box", "memorable", "not boring", "less dry", or "easier to consume"; whenever they push back with "this still reads the same", "too much t…

Audits copy that cites numbers, research, surveys or statistics against its sources, so every claim says no more than its source does. For each claim it quotes the source word for word, checks who was counted and the unit, labels the figure (measured, self-reported, survey, correlational, modelled, prediction, rule of thumb, vendor claim), dates it, then keeps, tightens or cuts it and rewrites it. Returns an audit table, a publishable "How to read these numbers" note and open checks, and applies the fixes to files when asked. Use before publishing a landing page, deck, report, blog post, press release, chart or app copy that quotes figures, and when the user says "fact-check this", "check these stats against the sources", "are these numbers right", "source-check the copy", "audit the claims", "does this overclaim", "tighten the wording to match the research" or "add sources and dates to the figures".

Extreme Pro README Writer for generating modern, visually stunning, and developer-friendly README.md files. Use this skill whenever the user wants to: create a README from scratch for any project, improve or modernize an existing README, rewrite documentation for a GitHub repo, generate a README from a codebase or project description, or make their project's first impression count. Also trigger when the user mentions 'README', 'repo documentation', 'project docs', 'GitHub landing page', wants to make their open-source project more appealing, or asks for help with any .md documentation file that serves as a project's main entry point. This skill covers all project types: mobile apps, backend services, SDKs, CLI tools, AI/ML projects, and web apps. Even if the user just says 'improve my docs' or 'make this look better' for a markdown file, use this skill.

Turn course material (slides, PDF, notes, syllabus, or just a topic) into a Duolingo-style interactive course as ONE offline HTML file, using a prebuilt tested engine so Claude only writes lessons. Use for "make a course", "duo course", "interactive course", "turn these slides into lessons".

Write or fix text for people so that it reads as written by this person, not by the average of the internet. Pipeline - philosophy (material and the reader first), a loop that pushes the draft away from what default models write for the same task (StoryScope-style discourse features with weights learned from the user's blind-test marks), then a surface lint. Any language. Use for letters, messages to hiring teams, cover letters, CV sections, bios, posts, READMEs, video scripts. Triggers - "не слоп", "ne-slop", "по-человечески", "живее", "звучит как ChatGPT", "sounds like AI", "unslop", "humanize", "fix this text", "write a message to".

Create learning editions of books from PDF, EPUB, or Markdown when readers need faithful meaning-based translation, brief explanatory notes, chapter summaries, or resumable processing of selected chapters or pages.