What Is Claude Code? The 2026 Guide for Founders and Senior Devs

Friday, 10 p.m. You open your terminal, type claude, paste the Linear ticket "fix the double-billing bug on Stripe", and go grab a coffee. When you get back, Claude has found the bug, fixed it across three files, gotten the tests passing and opened the PR with a clear message. You approve it. That's how Claude Code, Anthropic's coding agent, works in 2026: it lives in your terminal, reads your whole codebase, writes and tests code, then opens PRs for you. Anthropic also ships it as a VS Code and JetBrains extension, a desktop app and a web version, but its home turf is still the console you already have open. It runs on the current Claude models, including Opus 5.5 and Sonnet 5, and Anthropic includes it in the Pro plan at $17/month billed annually. Below we cover its features, release date, installation, the configuration that matters day to day, pricing plans, and how it compares with Cursor, GitHub Copilot and Codex.

What exactly is Claude Code?

Claude Code is an agentic coding tool built by Anthropic. Agentic means two specific things. First, it makes decisions without someone holding its hand. Second, it takes action: it reads files, runs shell commands, launches tests, edits code and makes Git commits. It does more than suggest.

The official docs sum it up as "an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools." Anthropic offers it on five surfaces connected to the same engine:

• the terminal, with the full CLI;
• VS Code, through an official extension that also works in Cursor;
• JetBrains IDEs (IntelliJ IDEA, PyCharm, WebStorm), through a plugin built on the CLI;
• the desktop app for macOS and Windows, in beta on Ubuntu and Debian, with a paid plan;
• the web at claude.ai/code, also accessible from the Claude app on iOS and Android.

Your CLAUDE.md files, settings and MCP servers follow you from one surface to the next. In Slack, mentioning @Claude on a bug report gets you back a pull request.

Claude Code release date and versions

Anthropic introduced Claude Code as a limited research preview in the Claude 3.7 Sonnet announcement, then opened it to everyone with the Claude 4 launch. Each new model generation then lands in the tool: the Claude 5 family (Opus 5.5, Sonnet 5, Fable 5.1) took over from Sonnet 4.6 and Opus 4.7, which Anthropic keeps available as legacy models.

On the software side, the native install updates itself in the background, and claude --version shows the installed version. Homebrew offers two channels: claude-code follows the stable channel, which runs about a week behind and skips releases with major regressions, while claude-code@latest gets every release as soon as it ships. With Homebrew or WinGet, you run updates yourself (brew upgrade claude-code, winget upgrade Anthropic.ClaudeCode).

How it differs from Claude.ai (the web version)

Claude.ai is the chat in your browser. You paste code into it, discuss, and copy back snippets of the answer. Everything goes through copy and paste. It's handy for brainstorming, but the context doesn't survive the session, and the tool never touches your project.

Claude Code is the opposite. You hand it a folder, and it reads the file tree, understands the dependencies and opens the files it considers relevant. It keeps your CLAUDE.md, your history and your conventions in mind, edits files in place and commits the result. It's the difference between a consultant you call on the phone and a senior dev you've hired full time.

If you want to place Claude Code within the new generation of tools where you code with AI rather than alongside it, the conceptual framework is vibe coding: you describe the intent, the agent executes.

For office tasks outside of code, like sorting files or producing a spreadsheet, Anthropic offers the same engine without a terminal in Claude Cowork. For mockups, prototypes and slides, Anthropic launched Claude Design.

To give it a persistent server with files, automations and hosted services, our Zo Computer review covers that execution layer in detail.

Claude Code's everyday tasks

The most common use cases:

• Reading an unfamiliar repo and explaining its architecture in 30 seconds
• Writing the missing tests for an untested module, running them, fixing the failures
• Refactoring a function across ten files in one go
• Reproducing a bug from an error message, finding the root cause, proposing a fix
• Creating a clean commit with a descriptive message and opening the PR
• Updating dependencies and resolving merge conflicts
• Translating a set of strings into ten languages through a CI script

You stay in your terminal and approve or reject each step.

Claude Code features

Here are the main features, each covered in detail in the section listed:

Install Claude Code and get started in 3 minutes

Installation takes about as long as a coffee. Anthropic recommends the native install, which creates a standalone binary that updates itself.

# macOS, Linux, WSL
curl -fsSL https://claude.ai/install.sh | bash

# Windows PowerShell
irm https://claude.ai/install.ps1 | iex

# Homebrew (macOS)
brew install --cask claude-code

# WinGet (Windows)
winget install Anthropic.ClaudeCode

# npm (legacy but still works)
npm install -g @anthropic-ai/claude-code

On native Windows, Anthropic recommends Git for Windows so Claude has Bash; without it, Claude Code falls back to PowerShell. On Linux, apt, dnf and apk can also install the tool. With npm, NEVER use sudo: it creates broken permissions that take more than an hour to sort out.

First launch and authentication

cd path/to/your/project
claude

Claude asks you to authenticate. You have three options: a Claude Pro or Max account (unified billing with claude.ai, recommended for solo devs), Claude Console via OAuth (pay-as-you-go through the API), or Bedrock / Vertex AI / Microsoft Foundry for teams working in restricted cloud environments.

The free Claude account doesn't give access to Claude Code. You need at least Pro.

Once it's running, type /help for the list of commands, /resume to pick up your last session, and /permissions to manage authorizations. You're ready to go.

CLAUDE.md: your project's permanent context

Installation tutorials often skip the CLAUDE.md file, yet it has a big impact on the quality of everyday use.

It's a markdown file you place at the root of your project. Claude Code reads it at the start of every session. Put in it everything you'd explain to a new dev joining the team: the stack, the conventions, the pitfalls, the commands that work and the ones that break things. There's no required syntax, just plain markdown.

Minimal example for an Astro / TypeScript stack

# CLAUDE.md

## Stack
- Astro 5 + Tailwind 4 + strict TypeScript
- Content in `content/fr/blog/` (markdown with frontmatter)
- Deploy with `npm run build && npm run deploy`

## Conventions
- No `any`: use `unknown` and then narrow
- All components in `.astro` (no React unless already present)
- Slug = kebab-case, never accented characters

## Useful commands
- `npm run lint` before every commit
- `npm run check` for Astro TypeScript checks
- Tests with `vitest run`

## Known pitfalls
- Never touch `content/_entry_example.md` (template)
- Images live on media.swanbase.co (CDN), not in the repo

Configuration levels

There are two places to put a CLAUDE.md: at the repo root (tracked in Git, shared with the team), and ~/.claude/CLAUDE.md at the user level (your personal preferences, favorite frameworks and style, applied to all your projects).

In 2026, the official docs added another layer: automatic memory. Claude Code learns as it works, saving the build commands that work, the pitfalls it runs into and the implicit conventions. It enriches your context on its own between sessions. If your repo already contains an AGENTS.md written for other coding agents, Claude Code reads it, either on its own or alongside the CLAUDE.md.

The llms.txt format follows the same logic for AI SEO: you document the terrain so agents can find their way around. We cover it in how to use llms.txt.

Plan Mode: think before you code

Plan Mode is the feature that separates good usage from industrial-scale accidents. You turn it on by pressing Shift+Tab twice (on Windows, sometimes Alt+M depending on the version).

The principle is simple. Claude switches to read-only mode. It explores your project, understands the request, and presents an implementation plan before making any change. You read the plan and correct it. Only then does it execute.

Why this matters: on a refactor that touches fifty files, an agent that charges ahead can neatly break your build and cost you a morning of cleanup. With Plan Mode, the agent lays out its strategy before writing the first line. If the approach is shaky, you catch it right there, before the damage is done.

It's also the mode to keep on for sensitive projects in production, finance or healthcare: a "look but don't touch" mode until you say "ok, go ahead."

A typical workflow: press Shift+Tab twice, explain the complex task ("add multi-tenant support to every API route"), let Claude propose a structured plan, critique and amend it, press Shift+Tab to return to normal mode, and Claude executes the approved plan.

Sub-agents, hooks, skills, plugins: the industrialization layer

Claude Code out of the box is powerful. Configured Claude Code is a different category of tool. The features below take you from solo use to team use.

Sub-agents: delegating to specialists

A sub-agent is a Claude Code inside Claude Code. It runs in its own context window, with its own system prompt, and has access to a subset of tools. The main agent coordinates, and the sub-agent carries out one specific task.

A typical setup: a code-reviewer agent that can only read and comment on code (never edit it), a test-writer that only writes tests, and a migrator that only touches database migration scripts. You separate responsibilities and limit how much can break.

You create a sub-agent via /agents, then "Create new agent". Storage: .claude/agents/ at the project level, ~/.claude/agents/ at the user level. Format: markdown with YAML frontmatter. You version them in Git, and your teammates get them automatically.

Hooks: event-triggered automation

A hook is a shell command that fires automatically at specific points in the Claude Code lifecycle. You don't have to ask for anything; it just runs.

The key events:

• PreToolUse: before a tool call (useful for blocking anything that touches .env or secrets/)
• PostToolUse: after one (perfect for prettier --write after every edit)
• Notification: when Claude is waiting for a response (to ping Slack or play a sound)

A concrete example: a PostToolUse hook that runs eslint --fix on every modified .ts file means you never see a PR with badly formatted code again. Configure it via /hooks.

Skills: packaging repeatable workflows

Skills (formerly custom commands) are recipes you write once and the whole team shares: /review-pr, /deploy-staging, /migrate-db. Each one is a mini-prompt with its own tools, rules and steps.

This is where you turn your Notion runbook.md into living automation. A new dev joins, types /onboard, and Claude sets up their environment, configures their keys and shows them the useful commands.

Plugins: the external catalog

Plugins extend Claude Code with skills, sub-agents, preconfigured hooks and MCP servers. You install them from a marketplace or by direct reference in the config.

MCP and Channels: connecting Claude to the rest of your stack

The Model Context Protocol (MCP) is an open standard for connecting AI tools to external data sources. Anthropic published it, and GitHub and OpenAI have adopted it.

An MCP server exposes tools that Claude Code can call: "read Linear ticket #234", "create a GitHub issue", "list the Slack messages that mention me". You install the MCP server, authorize it, and Claude drives the tool.

The ones that change a team's daily life: Linear (Claude reads and updates tickets), GitHub (search across PRs, issues and comments), Slack (reading the bug thread, posting updates), Google Drive (reading PRDs before building a feature), Notion (syncing internal docs), and your internal APIs through a fifty-line custom MCP.

Channels: Telegram, Discord, webhooks

Channels are a layer built on MCP that pushes external events into a running Claude Code session. A bug report comes in on Telegram, you forward it to your channel, and the session open on your machine picks up the context without any copy and paste. The support Discord, the founder's Telegram, the Stripe webhook for failed payments: everything converges on the dev session.

When we place Claude Code in a 2026 startup stack, we judge it on this integration dimension: an isolated tool is nice, but a tool connected to the whole stack has a multiplier effect. It's the same lens we apply in our review of AI tools for startups in 2026.

Case study: from Linear ticket to merged PR

The typical scenario: a Linear ticket saying "The Stripe webhook stopped processing refunds after the move to v2 of the API."

1. Fetch the ticket via MCP: /mcp linear get-issue ENG-432. Claude reads the summary, the comments and the assignee, with no back and forth between Linear and the terminal.

2. Switch to Plan Mode (Shift+Tab twice) and give the context: "Read src/webhooks/stripe.ts, identify what changed in v2, and propose a fix plan with tests." Claude spots that the v2 payload uses a new refund_reason field while the handler expects reason. It proposes adding the mapping, writing a test for both the old and new formats, and logging unknown payloads.

3. Approve the plan and press Shift+Tab to leave Plan Mode. Claude writes the code and runs the tests: both pass, covering both scenarios.

4. Say "Commit and open the PR." Claude writes a descriptive message, pushes, opens the PR on GitHub, and the Linear ticket moves to "In Review" automatically via MCP.

The ticket goes to review without you ever leaving the terminal. By hand, the same fix means bouncing between Linear, the IDE, GitHub and Slack.

Claude Code vs Cursor, Copilot, Codex in 2026

The question comes up at every demo. All four tools are good, and each one covers a different need. For an open source alternative that works with any model, read our OpenCode guide.

Cursor, Copilot and Codex pricing changes often, so check their official pages before deciding. For features specific to Claude Code (hooks, Channels, cloud routines), see the dedicated sections above.

Here's the short version for each tool.

Cursor remains the best choice if you live in your editor and want "VS Code with a copilot everywhere." If your day is spent clicking around an IDE, Cursor will suit you.

GitHub Copilot remains the standard for in-editor autocomplete, with an agent mode built on the GitHub ecosystem.

Codex CLI is OpenAI's terminal option, tied to the ChatGPT ecosystem.

Claude Code stands out on agentic work, multi-surface support (terminal, IDE, desktop, web, Slack, mobile), sub-agents and Plan Mode. It fits teams that want to industrialize the way they write code.

For a detailed head-to-head covering RAM, speed and pricing based on real usage, we have a dedicated comparison: Claude Code vs Cursor.

Pricing and plans in 2026

Anthropic lists its prices in US dollars. Here's the state of play as of September 26, 2026, taken from the official pricing page.

Claude subscription or Claude Console: which access to choose

There are two ways into Claude Code. A Claude subscription (Pro, Max, Team, Enterprise) covers claude.ai, the desktop app and Claude Code on a single bill: the first time you run claude, you sign in with your Claude account.

The Claude Console is for developers who prefer to pay as they go through the API. You create a key there, set it in the ANTHROPIC_API_KEY variable or sign in with your Console account, and Anthropic bills each token at API rates. In the terminal, VS Code and JetBrains, you can also go through Amazon Bedrock, Google Vertex AI or Microsoft Foundry. The desktop app requires a paid subscription.

API pricing per million tokens, taken from the models documentation:

With prompt caching, a cache read costs 10% of the base input price (5% on Opus 5.5, 2.5% on Fable 5.1). The difference adds up as soon as your CLAUDE.md grows.

The rule of thumb: Pro for 2 to 3 hours of coding a day, Max once you code full time, Team for a team that needs centralized administration, Enterprise for regulated industries, and the Console if you want to pay per token or plug Claude Code into a CI pipeline. The free plan doesn't include Claude Code, so there's no point looking for a way to enable it.

If your startup is less than five years old, the Claude for Startups program gives you one year of Claude Team for five Premium seats plus $1,000 in API credits.

FAQ

Is Claude Code free?

No. Anthropic includes Claude Code in the Pro plan at $17 per month billed annually or $20 billed monthly, as well as in Max, Team and Enterprise. The free Claude plan (claude.ai) doesn't give access to Claude Code. You can also pay as you go through the API with a Claude Console account.

Which models does Claude Code use in 2026?

Claude Code uses the current Claude family: Opus 5.5, which Anthropic recommends as the starting point and for agentic coding over long sessions, Sonnet 5 for the balance of speed and intelligence, Fable 5.1 for the most demanding reasoning, and Haiku 4.5 for fast operations. You can switch models mid-session.

When was Claude Code released?

Anthropic introduced Claude Code as a limited research preview alongside Claude 3.7 Sonnet, then opened it to everyone with the Claude 4 launch. The native install updates itself, and the claude --version command shows the installed version.

Can Claude Code modify my files without permission?

No. By default, Claude Code asks for permission before any file edit or shell command. Read-only operations go through automatically, while writes and Bash commands require approval, which you can make permanent for the session. The acceptEdits mode automatically accepts writes, and the plan mode blocks any change until you explicitly approve.

What's the difference between Claude Code and Claude.ai (web)?

Claude.ai is the chat interface in your browser: you paste code into it and work in question-and-answer mode. Claude Code is an agent that lives in your terminal, IDE or desktop: it reads your entire project, edits files, runs commands and makes Git commits. Claude.ai is an occasional consultant, while Claude Code is a developer embedded in your environment.

Can you use Claude Code in VS Code?

Yes. Anthropic publishes an official extension on the VS Code Marketplace that provides inline diffs, @-mentions, plan review and conversation history inside the editor. The same extension works in Cursor. JetBrains IDEs (IntelliJ, PyCharm, WebStorm) have their own plugin on the JetBrains Marketplace.

Is Claude Code safe for proprietary code?

Pro and Max plans aren't used to train models by default. Team and Enterprise add SSO, access controls, audit logging and advanced compliance. For regulated environments, you can route Claude Code through Amazon Bedrock, Google Vertex AI or Microsoft Foundry. With hooks, you can also programmatically block access to sensitive files.

Does Claude Code replace a developer?

No, and the question misses the point. With Claude Code, the developer moves from executor to supervising architect. The dev defines the intent, validates the approaches and reads the diffs, while the agent handles the mechanical writing, tests, commits and research. The effect is a multiplier.

Recommended articles

Claude Code vs Cursor: Which AI Coding Tool Should You Choose in 2026?

Claude Code and Cursor lead AI-assisted development in 2026. Claude Code is an autonomous agent that runs in your terminal, while Cursor is an IDE forked from VS Code. We cover SWE-bench results, euro pricing, a hybrid workflow and recommendations for each founder profile.

Mistral Vibe (formerly Le Chat): 2026 guide, pricing and Vibe Code

Mistral Vibe has been the new name for Le Chat since May 28, 2026, and it is also the name of Mistral AI's coding agent (CLI, VS Code extension, web). Pricing as of November 11, 2026, installation, models and a comparison with Claude Code.

Vibe coding: definition, how it works, tools and limits (2026)

Vibe coding means having an AI write an application's code from plain-language instructions.

Cursor, Lovable or Bolt: Which Vibe Coding Tool Should You Choose in 2026?

Lovable, Bolt.new or Cursor: a full comparison to help you pick the right vibe coding tool for your profile and startup stage. Pricing table included.

How to Find a CTO for Your Startup: The Guide for Non-Technical Founders (2026)

You have a startup idea, a validated market, maybe your first customers, and you don't code. Here's which technical profile fits your stage, where to find a CTO in France, how to evaluate one without coding, and how much equity to offer.

Claude Managed Agents: The Complete Guide to Deploying AI Agents in Production

Claude Managed Agents (Anthropic, April 2026) is cloud infrastructure for deploying AI agents in production without managing sandboxing, sessions, or permissions. It cuts months of development down to a few days.

Latest in this category

Claude Design: what it is, pricing, exports and use cases for founders (2026)

Claude Design as of November 2026: what Anthropic's tool produces, what it costs (included in every plan, Free too), its exports, its limits, and how it compares with Lovable, Claude Code, Canva and Figma.

OpenCode: What It Is, How to Install It, Go Pricing and Our Review (2026)

OpenCode, Anomaly's open source coding agent: what it does, how to install v2, which models you can connect, Go and Console pricing as of November 11, 2026, and how it compares with Claude Code, Codex and Mistral Vibe.

Supabase: what it is, what it costs and how to use it in 2026

Supabase in 2026: Postgres, auth and storage. Pricing as of Nov 11 (Free, Pro $25, Team $599), costs by stage, the MCP server and an RLS checklist before launch.