Claude Explained: How to Use Every Feature, and What's Actually Worth Paying For

Most people’s relationship with Claude starts and ends with a chat box: type a question, get an answer, done. That’s a fraction of what’s actually there. Underneath the chat interface sits a genuine work platform — one that can reach into your other apps, remember how you like things done, and carry out multi-step tasks with minimal hand-holding. This piece walks through what that platform actually contains: what Connectors and Skills are and how to use them to automate the boring parts of your week, what else is worth knowing about, and exactly where the free plan stops and paying starts making sense.

What Claude Actually Is Today

Claude is Anthropic’s AI assistant, available through a web app, desktop apps, and mobile apps for iOS and Android. Under the hood, several model families do the actual thinking: Claude Sonnet 5 and Claude Opus 5 are the current general-purpose and heavy-lifting models, Claude Haiku 4.5 is the fast and lightweight option, and Claude Fable 5 sits above Opus for long-running agentic work, alongside Claude Mythos 5 at the very top of the current lineup. Which model you can pick from depends on your plan, but the interface, the features, and the way you interact with Claude stay the same across all of them.

What’s changed over the last year isn’t really the chat experience — it’s everything built around it. Claude now creates and edits real files, executes code in a sandbox, browses the web, remembers context across separate conversations, and — through Connectors and Skills — reaches into the other tools you already use and applies your own specific way of doing things, automatically. That’s the part worth actually learning.

The Core Features Worth Knowing First

Before getting into Connectors and Skills, a quick tour of what’s sitting in Claude by default:

Artifacts — when you ask Claude to produce something substantial (a document, a piece of code, a chart, a webpage), it opens in a dedicated side panel instead of getting buried in the chat, so you can view, edit, and iterate on it directly.
Create and edit files with code execution — Claude can generate real Word docs, PowerPoint decks, Excel spreadsheets, CSVs, and PDFs, and run Python code in a sandboxed environment for genuine data analysis, not just describing what the analysis would show.
Projects — self-contained workspaces with their own chat history and a knowledge base of uploaded documents, plus custom instructions that apply to every conversation inside them. Free accounts can create up to five; every paid plan removes that cap entirely.
Memory — Claude retains relevant context across separate conversations, so you stop having to re-explain your role, your preferences, or an ongoing project every time you open a new chat.
Web search and extended thinking — both available from the free plan, for pulling in current information and working through more complex problems step by step before answering.
Voice mode, dictation, and incognito chats — for talking to Claude instead of typing, and for one-off conversations that don’t get saved to your history.

All of these are useful in isolation. The two features that actually change what Claude can do — rather than just how pleasant it is to use — are Connectors and Skills.

Connectors: Claude’s Bridge to Your Other Apps

A Connector links Claude to an external tool — Slack, Google Workspace, Notion, Asana, GitHub, a CRM, a database — so Claude can pull in real information from it and, where you allow it, take action inside it. They’re built on the Model Context Protocol (MCP), an open standard Anthropic created specifically so AI systems can plug into external tools and data in a consistent way. The Connectors directory currently lists several hundred of them, spanning categories like productivity, data, sales and marketing, code, design, financial services, and healthcare, built and maintained by both Anthropic and third-party developers.

How to add one

Open the “+” button in the lower-left corner of the chat box (or type “/” to bring up the same menu).
Select Connectors, which opens a directory browsable by category or searchable directly.
Pick a connector and follow its authentication flow — most use OAuth, so you’re logging into the actual service and granting specific, revocable permissions, not handing Claude a password.
In any conversation, make sure the connector is toggled on. Once it’s connected, Claude will often bring it into a conversation on its own when it’s clearly relevant, without you needing to name it every time.

If the tool you want isn’t in the directory, custom connectors let you point Claude at any remote MCP server directly — from Settings, choose to add a custom connector, paste the server’s URL, and it becomes available the same way a directory listing would. This is the option worth knowing about if your team has internal tools or a homegrown system you want Claude to work with.

A few things worth being deliberate about: Claude can only access what you’re personally permitted to see in the original service, all data transfer is encrypted, and connected services process your data on their own infrastructure under their own terms — so it’s worth a glance at a connector’s permissions before authorizing anything with write access to something sensitive. You can review and revoke any connector at any time from Settings, and it’s good practice to disconnect ones you’ve stopped actively using.

Skills: Teaching Claude Your Way of Working

If Connectors are about what Claude can reach, Skills are about how Claude does the work once it gets there. A Skill is a packaged set of instructions — at minimum, a folder with a SKILL.md file describing when and how to use it, often bundled with supporting templates, reference files, or scripts — that Claude loads automatically whenever it’s relevant, without you re-explaining your preferences every single time.

Here’s a simplified example of what a real Skill file looks like:

---
name: internal-comms
description: Resources for writing internal communications in our
  company's preferred format. Use for status reports, leadership
  updates, newsletters, FAQs, and incident reports.
---

## When to use this skill
Use this for 3P updates (Progress, Plans, Problems), company
newsletters, FAQ responses, status reports, and incident reports.

## How to use this skill
1. Identify the communication type from the request.
2. Load the matching guideline file from the examples/ directory.
3. Follow that file's instructions for tone, format, and structure.

Once that’s saved and enabled, you stop having to describe your company’s report format every time — you just ask for the report, and Claude already knows exactly how you want it.

Three kinds of Skills

Built-in Skills ship with Claude and activate automatically for common document work — creating a properly formatted Excel sheet with working formulas, a PowerPoint deck, converting between file formats — with zero setup.
Custom Skills are ones you build yourself, either by hand-writing a SKILL.md file or by describing what you want to the built-in skill-creator skill, which generates the folder structure and file for you.
Organization-provisioned Skills, available on Team and Enterprise plans, let an admin package a company’s procedures, brand guidelines, or compliance workflows and roll them out to every member automatically, so a new hire gets expert-consistent output from day one instead of everyone reinventing their own version of “the way we do things here.”

How to turn them on

For individual Free, Pro, and Max accounts: go to Settings → Capabilities, and make sure both Code execution and file creation and Skills are toggled on — Skills depend on code execution, so it won’t work without it. From there, Customize → Skills is where you enable built-in Skills, upload your own, or launch the skill-creator to build one from a description.

For Team and Enterprise: an Owner enables the same toggles under Organization settings → Skills, and can then upload a .zip containing a SKILL.md file to provision it for the whole organization, or restrict it to a specific group by bundling it into a plugin.

The genuinely useful part is that Skills work everywhere Claude does — claude.ai, Claude Code, Claude Cowork, and the API all draw from the same set, so a Skill you build once doesn’t need to be rebuilt for each surface. And because Claude decides which Skills apply to a given request on its own, you can stack several without manually picking one: ask for something that needs your brand voice, your data-analysis method, and your report format, and Claude pulls in whichever combination of Skills actually applies.

Putting Them Together: Automating Everyday Tasks

This is where Connectors and Skills stop being separate features and start functioning as an actual automation layer.

Weekly reporting, without the manual pull. Connect a data source or project-management tool, pair it with a Skill that knows your report format, and a single request (“pull this week’s numbers and write the Monday update”) replaces what used to be copying figures across five tabs.
Inbox and calendar triage. With Google Workspace or Outlook connected, Claude can summarize what actually needs a reply, draft responses in your voice using a writing-style Skill, and check them against your calendar before you’ve opened your inbox.
Client-ready documents on demand. A Skill holding your brand guidelines, combined with a Project containing a client’s brief and prior correspondence, turns “write the proposal” into a finished, correctly branded document instead of a rough draft you still have to reformat.
Team delegation through Slack. With @Claude (Claude Tag) set up in a workspace, any teammate can tag Claude into a channel and hand off a task directly — pulling in whatever Connectors and Skills are available to the workspace — without opening a separate app at all.
Background, multi-step work. For anything bigger than a single request — research that touches several tools, a task that should run on a schedule — Claude Cowork is built specifically for handing off longer jobs and letting Claude work through them with less supervision.

The pattern across all of these is the same: Connectors supply the raw access, Skills supply the judgment about how to use it, and Projects or Memory supply the ongoing context so you’re not re-explaining yourself in every single conversation.

The Rest of the Claude Ecosystem

A few other products worth knowing exist, since “Claude” now covers more than the chat window:

Claude Code — an agentic coding tool for the terminal, desktop, or mobile, for developers who want to delegate real coding work (writing, debugging, multi-file refactors) rather than copy-pasting snippets out of a chat.
Claude Cowork — a desktop app for handing off heavier, multi-step knowledge work — research, analysis, long-form writing — that benefits from Claude working across many tools and files with less back-and-forth.
Claude in Chrome — a browsing agent that can navigate and act on websites on your behalf.
Claude for Microsoft 365 (and standalone Claude for Excel / PowerPoint / Outlook) — brings Claude directly into Word, Excel, PowerPoint, and Outlook rather than requiring you to copy content back and forth.
@Claude (Claude Tag) — the Slack-based interface mentioned above, for tagging Claude into a channel like a teammate.
Claude Design and Claude Science — more specialized apps for prototyping/visual work and for research- and lab-oriented workflows, respectively.

What Requires a Paid Subscription, and What You Actually Get

This is the part worth being precise about, since a lot of casual coverage gets it wrong. Based on Anthropic’s current published pricing:

	Free	Pro	Max	Team	Enterprise
Price	$0	$17/mo (annual) or $20/mo	From $100/mo (5x) or $200/mo (20x)	$20–$100/seat/mo (annual)	$20/seat/mo + usage at API rates
Chat, Artifacts, web search, memory	Yes	Yes	Yes	Yes	Yes
Connectors	Yes	Yes	Yes	Yes, plus admin controls	Yes, plus admin controls
Skills	Yes	Yes	Yes	Yes, plus org-wide deployment	Yes, plus org-wide deployment
Projects	Up to 5	Unlimited	Unlimited	Unlimited, plus sharing	Unlimited, plus sharing
Claude Code, Cowork, Design, Science	No	Yes	Yes	Yes	Yes
Research	No	Yes	Yes	Yes	Yes
Claude for Microsoft 365	No	Yes	Yes	Yes	Yes
Relative usage	Base	5x+ more than Free	5x or 20x more than Pro	Varies by seat type	Usage-based, no fixed cap
SSO / SCIM / audit logs	No	No	No	SSO only	Full (SCIM, audit logs, compliance API)

A few things worth calling out that surprise people:

Connectors and Skills are not premium features. Both are available starting on the free plan, provided code execution is enabled — the thing that changes at higher tiers is administrative control (Team/Enterprise admins can govern which connectors are allowed org-wide) and organization-wide Skill deployment, not access to the features themselves.
What Pro actually buys you is less about unlocking a specific toggle and more about removing the ceiling: significantly more usage per session, no project cap, access to Research, and — importantly — Claude Code and Claude Cowork, neither of which are available on Free at all.
Max is a usage plan, not a features plan. Everything you get on Max, you also get on Pro; Max exists purely for people who’d otherwise be hitting usage limits constantly, trading a higher monthly price for meaningfully more headroom and higher output limits per task.
Team and Enterprise are organizational, not individual, upgrades. The jump isn’t about more personal capability — it’s central billing, SSO, admin-managed connectors, and (on Enterprise specifically) SCIM, audit logs, custom data retention, and a compliance API, all of which matter to IT and security teams far more than to a single user.
All plans, including Enterprise, exclude direct API usage from the flat fee. A Claude subscription covers claude.ai, the apps, and Claude Code; building your own product on top of the Claude API is billed separately, per token, through a developer account.

Frequently Asked Questions

Do I need a paid plan to use Connectors or Skills? No. Both are available on the free plan, as long as code execution is enabled in Settings. What paid plans add is administrative control and org-wide deployment, not access to the features themselves.

What’s the practical difference between Pro and Max? Max doesn’t unlock anything Pro doesn’t already have — it exists purely to give heavy users more usage headroom (5x or 20x Pro’s allowance) and higher output limits, for people who’d otherwise be waiting out session limits regularly.

Can I build my own Skill without knowing how to code? Yes — the built-in skill-creator skill will generate the folder structure and SKILL.md file for you based on a plain description of what you want, though writing one by hand gives you more control if you want it.

Is my data used to train Claude if I connect my tools? On paid plans, model training is opt-out by default; data accessed through Connectors is used to complete your request and isn’t retained beyond what that requires. If you’re on a Team or Enterprise plan, content isn’t used for training by default.

Do Skills and Connectors work in Claude Code and Cowork too, or just the chat app? Both work across claude.ai, Claude Code, Claude Cowork, and the API — a Skill or Connector you set up once doesn’t need to be rebuilt separately for each surface.

How many projects can I have on the free plan? Five. Every paid plan — Pro, Max, Team, and Enterprise — removes that cap entirely.

Final Verdict

Claude has quietly moved well past “a chatbot that writes decent emails.” Connectors and Skills, specifically, are the two features that turn it from something you talk to into something you can actually delegate to — and the fact that both are available on the free plan means there’s genuinely no reason not to try them, regardless of what you’re currently paying.

Where a subscription earns its cost is narrower and more honest than most coverage suggests: it’s not about “unlocking AI,” it’s about removing the usage ceiling and adding Claude Code, Cowork, Design, and Science to what you can reach for. If your work is occasional — a few writing or research tasks a week — the free plan, with a handful of Connectors and a couple of well-built Skills, will genuinely cover you. The moment Claude becomes part of your daily workflow rather than an occasional tool, particularly if that includes coding or longer autonomous tasks, Pro pays for itself quickly, and Max is there specifically for the point where even Pro starts to feel like it’s throttling you rather than helping you.

The honest recommendation: start free, connect the two or three tools you actually use every day, build one Skill around whatever task you repeat most often, and only pay once you can feel exactly where the free plan is holding you back — you will.

Scan Article

Claude Explained: How to Use Every Feature, and What’s Actually Worth Paying For

What Claude Actually Is Today

The Core Features Worth Knowing First