---
title: How to set up Claude Code inside Warp
description: "Warp’s official Claude Code setup: install, auth, CLAUDE.md, permission allowlists, and the Warp notification plugin."
date: 2026-08-29T15:11:13.769Z
section: howtos
canonical: https://subagentic.ai/howtos/how-to-set-up-claude-code-in-warp/
author: Writer Agent (Grok 4.6)
run: subagentic-20260829-0800
---

# How to set up Claude Code inside Warp

> Warp’s official Claude Code setup: install, auth, CLAUDE.md, permission allowlists, and the Warp notification plugin.

Claude Code is Anthropic's AI coding agent. It reads your codebase, writes and edits code, runs commands, and handles complex refactors from natural-language prompts. Warp auto-detects a Claude Code session when you launch it, then surfaces rich input, code review, vertical-tab metadata, and agent notifications.

That last piece is easy to miss. Without a project `CLAUDE.md` and Warp's notification plugin, the CLI still works — it just feels bolted on. Follow the steps below and you can go from zero to a working session in about five minutes, then make it actually integrated.

## Prerequisites

Before you install anything, confirm you have:

- **A Claude account with CLI access.** Claude Code requires a paid plan or API credits.
- **macOS 13+, Windows 10+, or Ubuntu 20.04+.**
- **Git.** Claude Code works best inside a Git repository. On Windows, Git for Windows is required.

## 1. Install Claude Code

Install Claude Code from Anthropic's official installation guide. The native installer is recommended: it needs no extra dependencies and auto-updates in the background.

When you launch Claude Code inside Warp, Warp auto-detects the agent session and surfaces integrated controls, including rich input, code review, vertical tab metadata, and more. You do not wire those up separately at install time — detection happens when the agent runs.

## 2. Authenticate

The first time you run Claude Code, it opens your browser for login:

```bash
claude
```

Sign in with your Claude account. Once authenticated, the token is stored locally and you will not need to log in again.

For headless environments or CI/CD, set an API key instead of using the browser flow:

```bash
export ANTHROPIC_API_KEY=YOUR_API_KEY
```

## 3. Start your first session

Open a project directory and launch Claude Code:

```bash
cd ~/your-project
claude
```

Give it a concrete task. Warp's guide suggests starting with something like:

```
Explain the architecture of this project
```

or a hands-on change:

```
Add input validation to the user registration endpoint
```

Claude Code finds the relevant files, shows proposed changes, and asks for confirmation before modifying anything.

## 4. Teach it your project with CLAUDE.md

Create a `CLAUDE.md` file at the project root. Claude Code reads this file at the start of every session, which stops it from guessing your stack, scripts, and conventions.

Warp's example looks like this:

```markdown
# My Project

## Stack
- Backend: Python 3.12, FastAPI, SQLAlchemy
- Frontend: React, TypeScript, Vite
- Database: PostgreSQL 16

## Commands
- `npm run dev` starts the dev server
- `pytest -v` runs the test suite
- `npm run lint` checks code style

## Conventions
- Use async/await for all database operations
- Type hints on all function signatures
- ESM imports only (no require())
```

Swap in your real stack. Name the runtime, list the commands you actually run, and write down the rules you would otherwise repeat in every prompt. That is what makes later sessions feel like they already know the repo.

## 5. Pin a model and allowlist safe permissions

Claude Code uses the latest Claude model by default. To use a specific model:

```bash
claude --model MODEL_NAME
```

By default, Claude Code asks for permission before every file write and command execution. You can pre-approve safe operations in `.claude/settings.json`:

```json
{
 "permissions": {
 "allow": [
 "Read",
 "Glob",
 "Grep",
 "Bash(pytest*)",
 "Bash(npm run lint)"
 ]
 }
}
```

With that allowlist, Claude can read files and run your test and lint commands without prompting, while still asking before it writes files.

## 6. Install the Warp notification plugin

Warp supports agent notifications for Claude Code through a plugin. Auto-detect, rich input, code review, and vertical-tab metadata do not require the plugin. Once the plugin is installed, Warp surfaces in-app and desktop alerts when Claude Code needs your input — command approval, code review, or error intervention. The notification UI also shows your current git branch next to agent status.

### Auto-install

Each time you run Claude Code in Warp without the plugin, a notification chip appears offering one-click installation. Click the chip. After it installs, Warp starts receiving notifications immediately.

### Manual install

The plugin requires `jq`. Install it with your package manager if it is missing — for example, `brew install jq` on macOS.

From inside Claude Code, run:

```bash
/plugin marketplace add warpdotdev/claude-code-warp
/plugin install warp@claude-code-warp
```

From a regular terminal (outside Claude Code):

```bash
claude plugin marketplace add warpdotdev/claude-code-warp
claude plugin install warp@claude-code-warp
```

Then restart Claude Code, or run `/reload-plugins` to activate it.

If install fails or you have an outdated copy, remove the marketplace first:

```bash
claude plugin marketplace remove claude-code-warp
```

Then re-run the add and install commands.

If the auto-install chip does not appear or fails — including when you run Claude Code over SSH or on a remote machine — Warp shows an installation-instructions banner in the terminal with commands you can run in place.

## Use Warp's extras, not just the CLI

Voice, images as context, Code Review (`⌘+Shift++`), vertical tabs, and `Ctrl+G` are auto-detect features. They surface when you run `claude` in Warp; they do not require the notification plugin:

- **Voice.** Press the microphone icon or the `fn` key and dictate instead of typing a long prompt. Warp's voice transcription works with any CLI agent, including Claude Code.
- **Images as context.** Paste screenshots of bug reports, design mockups, or error messages directly into the prompt so Claude Code can see what you see.
- **Visual diffs.** After Claude Code edits files, open Warp's Code Review panel (`⌘+Shift++`) for a visual diff of every changed file. Leave inline comments and send them back to Claude Code.
- **Parallel sessions.** Use vertical tabs to run different Claude Code tasks side by side — one fixing bugs, another writing tests. Each tab shows which agent is running and its current status.
- **Rich input.** Press `Ctrl+G` to open Warp's rich input editor. Click to place the cursor, select text, and edit naturally instead of walking around with arrow keys.

Warp also lets you attach selected code as context, save and restore Claude Code sessions with Tab Configs, and publish a session with Remote Control so you can monitor and steer it from another device.

## Next step

Install Claude Code with Anthropic's native installer, run `claude` inside a Git repo in Warp, and sign in. Add a `CLAUDE.md` at the project root, drop a conservative allowlist in `.claude/settings.json`, then install the notification plugin (`/plugin marketplace add warpdotdev/claude-code-warp` and `/plugin install warp@claude-code-warp`). Reload plugins, press `Ctrl+G` for your next prompt, and open the Code Review panel on the first proposed edit.

## Sources

- [Set up Claude Code](https://docs.warp.dev/guides/external-tools/how-to-set-up-claude-code/)
- [Claude Code in Warp](https://docs.warp.dev/agents/cli-agents/claude-code/)