Word documents
Creates, reads and edits Word documents, including tracked changes, comments and formatting.
A curated library of Claude Skills for documents, data, design, code, research, business, teaching and games. Read every SKILL.md, stack the ones you need, and install them in a couple of minutes.
No sign-up and no card. The library moves to a paid plan after the launch window, so stack what you need while it is open.
Official Anthropic skills sit next to GeeklyPlug originals. Open any card for its SKILL.md, folder layout and install command.
Creates, reads and edits Word documents, including tracked changes, comments and formatting.
Reads, fills, merges, splits and creates PDF files, including form fields and text extraction.
Builds and edits PowerPoint decks with layouts, charts and speaker notes.
Creates and edits spreadsheets with formulas, formatting and charts, and recalculates them.
Turns meeting transcripts or rough notes into clean minutes with decisions, action items, owners and due dates.
Summarizes contracts in plain language and flags clauses on renewal, termination, liability, payment and data.
Cleans messy CSV files by fixing headers, types, dates, duplicates and blank rows, and reports every change.
Explains what a SQL query does in plain language, then suggests safer and faster versions.
Builds a single-file HTML dashboard from a CSV or spreadsheet with KPI tiles, trend charts and filters.
Codes open-ended survey answers into themes with counts and example quotes.
Guides distinctive front-end design so interfaces avoid generic, templated looks.
Creates posters and visual art as PNG or PDF from a written design philosophy.
Makes generative art with p5.js using seeded randomness and tunable parameters.
Applies a ready-made color and font theme to slides, docs, reports and pages.
Creates animated GIFs sized and tuned for Slack.
Audits HTML pages and components against WCAG 2.2 AA and returns fixes ranked by impact.
Guides building MCP servers that connect Claude to outside APIs and services, in Python or TypeScript.
Tests local web apps with Playwright: drives the UI, captures screenshots and reads browser logs.
Reviews diffs and pull requests for bugs, security issues, tests and readability, with comments tied to line numbers.
Writes Conventional Commit messages from staged changes.
Turns OpenAPI specs or route code into readable API reference docs with examples.
Fact-checks claims in a draft against primary sources and marks each as supported, disputed or unverified.
Builds a literature matrix from papers with question, method, sample, findings and limits side by side.
Breaks down a competitor's product, pricing, positioning and messaging into a one-page teardown.
Edits copy to match a house style guide: tone, banned words, punctuation and formatting.
Creates SEO content briefs with search intent, outline, questions to answer and internal links.
Assembles a newsletter issue from links and notes with an intro, sections, summaries and a subject line.
Walks a user through co-writing documentation, proposals and specs in structured rounds.
Writes internal communications such as status reports, leadership updates, FAQs and incident reports in company formats.
Writes short cold email sequences personalized from a prospect's site and role.
Turns customer interviews and results into a case study with challenge, solution, results and a pull quote.
Turns a messy process description into a step by step SOP with roles, inputs and checks.
Runs a weekly review covering wins, misses, open loops and next week's top three.
Creates quizzes from notes or chapters with plausible distractors, answers and explanations.
Builds lesson plans with objectives, timing, activities, checks for understanding and homework.
Explains a concept at three levels: for a 12 year old, a student and an expert.
Turns changelogs and commit lists into player-friendly patch notes grouped as new, balance, fixes and known issues.
Drafts game design documents covering pillars, core loop, mechanics, progression and scope.
Helps create, test and improve skills, including evals and description tuning.
Reviews a SKILL.md for clear triggers, scope, length and overlap with other skills.
Skills use progressive disclosure. Claude reads a short line about every skill, then opens the full instructions only when a task matches. Tap a layer to see where it lives in the folder.
Layer 1 lives in the frontmatter at the top of SKILL.md. Claude keeps this line for every installed skill in mind and uses it to decide what to open.
Fill in the fields and the file writes itself. The checks are the same ones we run before a skill goes into the library.
--- name: invoice-chaser description: Drafts polite, firm reminder emails for overdue invoices with the amount, due date and payment link. Use when the user mentions an unpaid or overdue invoice or asks to chase a payment. --- # Invoice Chaser ## When to use - Any request that matches the description above ## Steps 1. Find the client name, invoice number, amount and due date. 2. Pick the tone from days overdue: friendly under 14, firm after 30. 3. Write a subject line and an email under 120 words with the payment link. 4. Offer a follow up date. ## Rules - Never threaten legal action unless the user asks. - Always restate the exact amount and invoice number.
A skill is a folder with a SKILL.md inside. Where you put that folder depends on how you use Claude. Menu names change over time, so check Anthropic's docs if a label looks different.
Open Settings in the Claude app and find the Skills section. Code execution needs to be on, since many skills run scripts.
Zip the skill folder so the SKILL.md sits inside a folder named after the skill, like meeting-minutes/SKILL.md.
Upload the zip in the Skills section. Anthropic's official skills are already listed there to switch on.
Ask for the task in normal words. Claude reads the description and loads the skill when it matches.
A folder of instructions, and optionally scripts and reference files, that teaches Claude how to do one kind of task your way. Claude loads it on its own when your request matches the skill's description, so you stop pasting the same long prompt every time.
Yes, for a limited time every skill in the library, including the full SKILL.md files, the builder and the install script, is free to use with no sign-up.
There is no small practical cap. An idle skill costs only its name and description, roughly 100 tokens. The real limit is clarity: when two descriptions overlap, Claude can pick the wrong one, so keep each skill focused on one job.
Almost always the description. It needs the words people actually type, like slides, deck or invoice, and a clear use when clause. Run it through the builder on this page or the skill-auditor in the library.
Only install skills you trust or have read. A skill can include scripts that Claude runs, so treat a skill like any other code: read the SKILL.md and every script before you add it. Every GeeklyPlug skill here shows its full SKILL.md before you install.
Custom instructions and project knowledge are always in context. A skill sits idle until a task needs it, can carry code and files, and works the same way across the Claude app, Claude Code and the API.
Cards marked Anthropic point to Anthropic's public skills repository on GitHub. We show a summary here and link to the real files, since that repo is the source of truth.
We write, test and tune custom skills for teams: your formats, your tone, your tools. Every skill ships with trigger tests and a plain-English readme.