Hey, it's Cindy 🌱 You commented STACKED, so here are the three free resources from the reel, two of them from Anthropic.
btw the waitlist for AI System Club is now open! If you're keen on using AI to automate your content, video editing, job search, business or build AI agents, feel free to sign up (only if you want to hehe). Ps. the waitlist gets first dibs:)
the shape of it
Why these three stack
Most people collect resources and never build anything with them. These three are worth doing together because each one solves a different problem. Anthropic Academy teaches you the moves. The official skills library gives Claude reusable instructions so it performs one job properly every time. The community subagents repo gives you a whole bench of specialists you can hand work to. Learn, then install, then build.
1. Anthropic Academy Free official courses with certificates. This is where the fundamentals live: prompting, Claude Code, agents, MCP.
2. Official skills library Anthropic's public repo of Agent Skills. Drop one in and Claude suddenly handles slide decks, spreadsheets or design systems properly.
3. Awesome Claude Code subagents A community repo of 158+ prebuilt Claude Code specialists, from code reviewers to product writers.
The stack effect Academy gives you the vocabulary, skills give Claude the process, subagents give you the team. Doing one alone is where most people stall.
resource one
Anthropic Academy
Anthropic publishes a full course library at anthropic.com/learn, with the catalogue hosted at academy.claude.com. Courses are free to register (the enrol button on a course page literally reads "Register | FREE") and each one ends with a certificate of completion you can put on LinkedIn.
There are more than 20 courses in there, which is exactly why people open it once and close it again. Here is the order I would do them in.
🌱 Claude 101 Start here even if you have used Claude for months. It covers projects, artifacts and skills, plus connecting Claude to your other tools, which is the part most self-taught users have never touched. Open Claude 101
💡 AI Fluency: Framework & Foundations The thinking course rather than the tool course. It gives you the vocabulary to describe what you are actually asking a model to do, which quietly makes every prompt you write after it better. Open AI Fluency
👩💻 Claude Code 101, then Claude Code in Action 101 gets you installed and oriented. "In Action" is the longer build course where you work through real tasks in a codebase. Do them back to back. Claude Code 101 · Claude Code in Action
🧠 Introduction to agent skills This is the course that pairs with resource two below. Do it right before you install your first skill so the concepts land while you are actually using them. Open the skills course
🧠 Introduction to subagents And this one pairs with resource three. Subagents are the concept people find hardest to picture, so learning it before you install fifteen of them saves a lot of confusion. Open the subagents course
💡 Introduction to Model Context Protocol Optional, and worth it the moment you want Claude reading your real tools rather than pasted text. MCP is how Claude connects to things like your calendar, your database or your ad account. Open the MCP course
If you have one evening, do Claude 101 and AI Fluency. That pairing takes you from "I type things into a chat box" to "I know what I am configuring and why," and everything else in this guide gets easier afterwards.
resource two
Anthropic's official skills library
A skill is a folder with a SKILL.md inside it: a name, a one-line description of when to use it, and a set of instructions. Claude reads the description automatically and pulls in the full skill the moment your request matches. Anthropic keeps a public repo of the ones they build at github.com/anthropics/skills, and it is one of the most starred repos on GitHub.
These are the ones I would install first, based on what people actually ask Claude to do.
skill-creator The one that makes the others. It interviews you about a task you repeat, then writes and packages a custom skill for you. Start here.
pptx, docx, xlsx, pdf The document skills. They are why Claude can hand you a real formatted deck, doc or spreadsheet instead of a wall of text you have to rebuild.
brand-guidelines Teach Claude your fonts, colours and tone once, and every asset it makes afterwards comes out on brand instead of generic.
frontend-design The taste layer for anything you build. It stops vibe-coded pages from having that default AI-purple-gradient look.
canvas-design For visual layouts and design work where you want Claude composing on a canvas rather than describing.
mcp-builder For when you are ready to connect Claude to a tool that does not have a connector yet. It helps you build the bridge.
Installing one takes about a minute either way you do it.
💻 In the Claude app or claude.ai Download the skill folder from the repo and zip it, then open Customize, go to Skills, press Add, and choose Upload a skill. Custom skills work on every plan, Free included. Anthropic's walkthrough is here: using skills in Claude.
👩💻 In Claude Code Run /plugin marketplace add anthropics/skills, then /plugin install document-skills@anthropic-agent-skills or /plugin install example-skills@anthropic-agent-skills. You can also drop a skill folder straight into ~/.claude/skills/.
🌱 Test it on one real task Ask for the thing the skill covers, in normal language, and watch whether Claude reaches for it on its own. If it does not trigger, the description line in the SKILL.md is too vague, so make it say exactly when to use it.
Skills can run real instructions and scripts, so treat them like software: install from Anthropic's repo or from people you actually trust, and read the SKILL.md before you add anything you found on the internet.
resource three
Awesome Claude Code subagents
A subagent is a specialist Claude Code can hand a piece of work to, with its own instructions and its own context window. The community repo at github.com/VoltAgent/awesome-claude-code-subagents collects 158+ of them across ten categories, from core development and language specialists through to business, product and research roles. It is MIT licensed, so you can read and edit any of them.
code-reviewer Reads your changes with fresh eyes before you ship. The single highest-value one to install if you build anything.
security-auditor Checks for the obvious holes: exposed keys, unsafe inputs, permissions that are wider than they need to be.
frontend-developer Handles the interface work while your main thread keeps the plan. Useful the moment a build gets bigger than one file.
product-manager One of the business-category agents. Turns a vague idea into scoped requirements you can hand to the build agents.
📥 Install the ones you want Copy an agent's markdown file into ~/.claude/agents/ to use it everywhere, or into .claude/agents/ inside a project to keep it scoped to that project. The repo also ships a plugin route: claude plugin marketplace add VoltAgent/awesome-claude-code-subagents.
🎯 Start with two, not twenty Every installed agent adds a description Claude has to consider. Two well-chosen specialists beat a folder of forty you never call.
✍️ Edit them into your world These are plain markdown files. Open one, rewrite the instructions to match your stack, your tone and your standards, and it stops being generic.
copy these
Two prompts to make it stick
Reading a course list does nothing on its own. Paste this into Claude to turn the three resources into a plan that fits your actual week.
💡 build my two-week Claude plan
I want to properly learn Claude over the next two weeks using only free official resources, and I have about [X hours] a week. My job is [describe your role and the work you repeat most], and the outcome I want is [what you want to be able to do at the end, for example "turn my weekly reporting into something Claude does for me"].
Build me a two-week plan using these three resources: Anthropic Academy at academy.claude.com, the official skills library at github.com/anthropics/skills, and the community subagents repo at github.com/VoltAgent/awesome-claude-code-subagents. For each session tell me what to study, roughly how long it takes, and one small thing to build or install straight afterwards so the learning sticks.
Bias the plan towards building. If a session is more than half reading, replace part of it with a task from my real work. At the end of the two weeks I should have at least one custom skill and one subagent I actually use.
And once you have done the skills course, this is the prompt that turns your most annoying repeated task into a skill you keep.
🛠 turn my repeated task into a skill
I want to build a reusable Claude skill for a task I do every week: [describe the task in detail, including the inputs you start with and the output you need].
First ask me the three or four questions you need answered about my format, tone and edge cases. Then write me the complete SKILL.md: a short lowercase-hyphenated name, a description line that says both what it does and exactly when to use it, and clear step-by-step instructions written the way I would explain them to a new hire.
Then tell me how to save it as a zipped folder I can upload, and give me two test requests I can run to check the skill triggers on its own rather than me having to name it.
The description line is the part people get wrong. "Writes reports" is too vague for Claude to reach for it. "Use when I paste raw meeting notes and ask for the Monday client update" triggers reliably.
the honest version
What this does and does not give you
Anthropic Academy certificates are certificates of completion, not an industry credential, and nobody is hiring on the strength of one alone. What they are genuinely good for is proof of momentum: a dated, verifiable trail that you learned the thing properly and from the source. The real leverage is the second half of this guide, because a skill you built for your own job and a subagent you edited to match your standards are things you can demo, and a demo beats a badge every time.