Claude Code Plugins (2026): What They Are and the Cost
Claude Code plugins are one extension point among several, and the honest starting question is not what they can do but which plan you have to be on to find out.
Where plugins sit in the product line
Anthropic's own site groups its capabilities into a short list: Artifacts, Design, Connectors, Plugins and Skills, with extensions for Chrome and Microsoft 365 listed separately. Claude Code appears under Products and again under Build with, next to the Claude Platform. The practical reading is that plugins are a way of extending the assistant rather than a standalone marketplace product, and that the detail lives in the developer documentation rather than on the marketing pages. That matters when you are evaluating: the homepage will tell you plugins exist and which plan includes Claude Code, but the mechanics, how you install one, what it can reach, how it is scoped, are a documentation question, and worth reading before you build anything on top of them.
What the plans cost
Pricing is published openly. Free is $0 and covers chat across web, iOS, Android and desktop, code generation, web search, memory across conversations, file creation and code execution, connectors over remote MCP, and extended thinking. Pro is listed at $17 per month on the annual plan, billed as $200 up front, or $20 billed monthly, and that is the tier where Claude Code, Cowork, Design and Science are included, along with more usage and access to more models. Max starts at $100 per month and offers a choice of five or twenty times the Pro usage, higher output limits, earlier access to new features and priority at busy times. Anthropic notes that usage limits apply and that prices and plans can change.
What the homepage does not settle
Two things are left to the documentation, and both are the ones that decide a real project. First, the operational detail of the plugin system itself: how plugins are distributed, what permissions they carry and how they behave in a team setting. Second, the shape of the usage limits. The pricing table describes Pro as more usage and Max as five or twenty times that, without publishing the unit, so the only way to know whether your workload fits is to run it. Anyone budgeting for heavy automated use should read the linked usage-limit guidance before assuming a tier, and should treat the first month as a measurement exercise rather than a commitment.
When you do not need an agent at all
Agents and their plugins pay off on codebases you return to: repeated context, project conventions, long-running work. They are a poor fit for a one-shot deliverable. If the task is a landing page, a small marketing site, or a mobile app shell you want to hand off as files, the setup cost of a terminal agent and its extensions is larger than the task. Begin.sh takes a prompt, or a URL to clone, and gives you a working static site or Expo app as a zip. No hosting, no backend, no auth, which is both its limitation and the reason it finishes in minutes.
What you actually get
Bundled, not sold separately
Claude Code is listed as part of Pro rather than as its own purchase, so the plugin question is really a subscription question. Free covers chat and code generation but not Claude Code.
One of several extension points
Plugins appear next to Skills and Connectors in the capability list. Before building one, check which of the three is the intended fit for what you are trying to add.
Usage is the variable
Every tier is described in relative usage terms rather than absolute units. The tier you need depends on your workload, which you can only find out by running it for a while.
Documentation is the real source
Marketing pages confirm plugins exist and what the plans cost. Mechanics live in the developer docs, and that is where an evaluation should actually start.
Agent workflow or one-shot build
| Feature | Claude Code plugins | Begin.sh |
|---|---|---|
| What it is | Extensions inside a coding agent | A generator that hands you a project |
| Entry cost | Included with Pro at $17 a month annually, $20 monthly | See begin.sh for current terms |
| Setup | Install the tooling, configure, then extend | Type a prompt or paste a URL to clone |
| Output | Changes in your own repository | A downloadable zip: static site or Expo app |
| Backend and auth | Whatever you build yourself | Not included, by design |
| Good for | Ongoing work in a codebase you own | A site or app shell you need finished today |
Evaluating this in an afternoon
- Read the capability list first
Check whether your idea is a plugin, a skill or a connector. Building the wrong one is the most common way to waste a week on an otherwise sound integration. - Price the plan honestly
Pro is the tier that includes Claude Code, at $17 monthly on the annual plan or $20 month to month. Decide whether that is a tool budget or a personal one before you subscribe. - Measure your usage
Run a normal week of real work before upgrading to Max. The published tiers are relative, so your own pattern is the only meaningful input to that decision. - Ask if the task even needs it
For a single site or app handoff, a generator finishes faster than configuring an agent. Keep both routes open and choose per project rather than by habit.
FAQ
Do Claude Code plugins cost extra?
Nothing on the pricing page lists a separate charge for them. Claude Code is described as included with Pro, which is $17 per month on an annual plan billed at $200 up front, or $20 billed monthly. Max starts at $100 and provides five or twenty times the Pro usage. Usage limits apply and Anthropic states plans may change.
Can I use Claude Code plugins on the free plan?
The free tier is described as chat across web, mobile and desktop with code generation, web search, memory, file creation and code execution, plus connectors through remote MCP. Claude Code itself is listed under Pro rather than Free, so the practical answer is that the paid tier is where this sits. Check the current pricing page before relying on that.
What is the difference between plugins and skills?
Both are listed as capabilities on the same page, which is exactly why the distinction is worth reading properly in the developer documentation rather than inferring from a marketing list. Picking the wrong extension model is a costly mistake to discover late, so spend the ten minutes on the docs before you start building against either.
Is there a usage limit I should worry about?
The pricing page repeatedly notes that usage limits apply and links to guidance on working within them, but publishes tiers in relative terms rather than absolute units. If you plan to run automated or long-running work, assume limits will matter, measure a typical week on the lower tier, and upgrade based on what you observed rather than on an estimate.
I just want a website built. Is this the right tool?
Probably not the fastest route. An agent plus plugins is built for continuing work in a codebase. Begin.sh takes a prompt, or a URL you want cloned, and returns a working static site or Expo app as a zip you can download. It deliberately skips hosting, backend and auth, which is why it finishes quickly.
Which models come with the subscription?
The site lists Mythos, Fable, Opus, Sonnet and Haiku as the model family, and describes Pro as giving the ability to use more Claude models than Free, with Max adding early access to advanced features and priority during busy periods. Exactly which models map to which tier is worth confirming on the pricing page, since that mapping changes over time.
Skip the setup when you only need the site
Begin.sh turns a prompt, or a URL you want cloned, into a working static site or Expo app and hands you the zip. No hosting to configure, no agent to install.
Try Begin.sh →