Reference · Reference

Claude Code Features Guide

Reference guide to Claude Code features. Browse by category — code editing, Git, testing, MCP, customization — with search and filtering.

Last updated: September 5, 2026

How to Use

Expand how to use

1
Search for a Feature
Type a keyword in the search box or filter by category (Code Editing, Git, etc.) to find the feature you need.
2
View Feature Details
Check the description, usage examples, and related commands for each feature.
3
Try It Out
Use the examples as a reference and try the feature in Claude Code.

Recent Updates

Check the latest Claude Code features

v2.1.261 (September 4, 2026)

Added /skill-doctor to show what each skill costs in context and how often it is used, so you can find skills to turn off, an extension of the existing skills feature rather than a standalone feature
Added --append-subagent-system-prompt-file to read the subagent system prompt append text from a file, for prompts too large to pass on the command line, an extension of the existing subagent features rather than a standalone feature
Added bashOutputMaxChars and taskOutputMaxChars settings to raise how much command and background-task output Claude receives inline before it is saved to a file, up to 128K characters, a setting on the existing settings feature rather than a standalone feature

v2.1.260 (September 3, 2026)

Added a diff panel that opens beside the conversation in fullscreen mode and shows your uncommitted changes as Claude edits; toggle it with /diff
Added a likely cause for prompt-cache misses (e.g. tool definitions or system prompt changed, idle past the TTL) to /cost and the status line's prompt_cache field
Added /reload-plugins to headless sessions, so it appears in the Claude Code Desktop and SDK command lists

View full update history

71 features

File Read & Write

Read, create, edit, and overwrite files in your project. Supports reading multiple files simultaneously.

Code Editing

Usage Examples

Read src/app/page.tsx

Create a new utility function in utils/helpers.ts

Multi-File Editing

Edit multiple files across your project in a single instruction. Ideal for interface changes that require updates in related files.

Code Editing

Usage Examples

Add an email field to the User interface and update all related files

Unify all import paths to use @/ across all components

Code Generation

Generate code from natural language instructions. Supports functions, components, API endpoints, configuration files, and more.

Code Editing

Usage Examples

Create an API endpoint for sending emails

Implement an authentication middleware

Refactoring

Improve code structure, rename symbols, and apply design patterns while preserving behavior.

Code Editing

Usage Examples

Extract this logic into a custom hook

Convert this class component to a function component

Bug Fixing

Analyze error messages and stack traces to identify root causes and implement fixes.

Code Editing

Usage Examples

Fix this error: TypeError: Cannot read properties of undefined

Debug why login returns a 500 error and fix it

Code Review

Review code changes for bugs, security risks, performance issues, and best practice violations.

Code Editing

Usage Examples

Review the latest changes

Review this PR from a security perspective

Related Commands

Auto Lint Fix

Automatically detect and fix ESLint, Prettier, and other linting errors. Instantly fixes violations after code edits.

Code Editing

Usage Examples

Fix all lint errors

Apply Prettier formatting

Commit Creation

Analyze changes and auto-generate appropriate commit messages. Supports Conventional Commits format.

Git Operations

Usage Examples

Commit the changes

Commit with message: feat: add user authentication

Related Commands

Pull Request Creation

Automatically create PRs with change summaries and test plans using GitHub CLI.

Git Operations

Usage Examples

Create a PR

Create a PR targeting the develop branch

Related Commands

PR Review

Read and review existing PR changes, providing review comments.

Git Operations

Usage Examples

Review PR #123

Check this PR and flag any issues

Related Commands

Branch Management

Create, switch, merge, and delete branches.

Git Operations

Usage Examples

Create and switch to feature/auth-refactor branch

Merge latest from develop branch

Diff Analysis

Analyze git diff results and explain change summaries and impact scope.

Git Operations

Usage Examples

Show diff against main branch

Summarize changes since last commit

Related Commands

Conflict Resolution

Detect merge conflicts and resolve them by understanding the intent of both changes.

Git Operations

Usage Examples

Resolve merge conflicts

Fix conflicts during rebase

History Investigation

Investigate code change history using commit logs and git blame.

Git Operations

Usage Examples

Show the change history of this file

Find when this function was added

Related Commands

Command Execution

Execute shell commands and analyze results. Supports builds, tests, deployments, and any CLI operation.

Terminal

Usage Examples

Run npm run build

Check Docker container status

Related Commands

Environment Setup

Set up development environments, install dependencies, and generate configuration files.

Terminal

Usage Examples

Set up the development environment for this project

Create a .env file template

Related Commands

Package Management

Install, update, and audit packages with npm/yarn/pnpm.

Terminal

Usage Examples

Install zod

Run npm audit and fix vulnerabilities

Build & Run

Build projects, start development servers, and verify production builds.

Terminal

Usage Examples

Start the development server

Verify the production build passes

Process Management

Start/stop background processes and check port usage.

Terminal

Usage Examples

Check what's using port 3000

Start the dev server in the background

Prompt Suggestions

Auto-suggest next prompts based on git history and conversation context. Press Tab to accept or Enter to accept and submit. Cost-efficient background generation.

Terminal

Usage Examples

Press Tab to accept a suggestion

Press Enter to accept and submit immediately

/btw (Side Questions)

Ask quick questions about your current work without adding to conversation history. Available even while Claude is processing a response.

Terminal

Usage Examples

/btw what was the name of that config file again?

/btw what's the type of this function's argument?

Related Commands

Session Recap

One-line summary of what happened in the session, shown automatically after you step away and return. Also available on demand via /recap. Reuses the parent conversation's prompt cache for minimal cost.

Terminal

Usage Examples

/recap to generate a summary of the current session

Automatic recap appears after the terminal is unfocused for 3+ minutes

Related Commands

Codebase Search

Search by file name patterns, code content, and advanced regex patterns.

Project Understanding

Usage Examples

Find all files related to authentication

List all components that use useEffect

Related Commands

Dependency Analysis

Analyze import relationships between files, package dependency trees, and detect circular dependencies.

Project Understanding

Usage Examples

Visualize this module's dependencies

Check for circular dependencies

Code Explanation

Explain complex code behavior, algorithms, and design intent in clear terms.

Project Understanding

Usage Examples

Explain what this regex does

Explain how this custom hook works

Architecture Analysis

Analyze overall project structure, layer composition, and design patterns.

Project Understanding

Usage Examples

Explain this project's architecture

Suggest improvements to the directory structure

Impact Analysis

Identify the scope of code changes and list related files that need modification.

Project Understanding

Usage Examples

Check what's affected if I change this API type definition

Verify if it's safe to delete this function

Image Input (Multimodal)

Accept screenshots, design mocks, and error screens as input. Convert images to code or identify issues. Paste from clipboard with Ctrl+V.

Project Understanding

Usage Examples

Convert this design mock image to a React component

Ctrl+V to paste an error screenshot and identify the cause

Test Generation

Auto-generate test code for functions and components. Covers happy paths, error cases, and edge cases.

Testing

Usage Examples

Write unit tests for the calculateTotal function

Create tests for the LoginForm component

Test Execution & Debugging

Run tests and analyze/fix failing test causes.

Testing

Usage Examples

Run all tests

Debug and fix the failing tests

Test-Driven Development (TDD)

Follow the RED-GREEN-REFACTOR TDD cycle — write tests first, then implement.

Testing

Usage Examples

Implement user validation using TDD

Write the tests first, then implement

Test Maintenance

Update existing tests for implementation changes and improve test coverage.

Testing

Usage Examples

Update tests to match the API spec changes

Add tests where coverage is below 80%

Run & Verify Your App

Launch and drive your app to confirm a change works in the running app, not just in tests or type checks. Built from the /run, /verify, and /run-skill-generator bundled skills (v2.1.145+).

Testing

Usage Examples

Run the app and check that my change works

Build and run it to verify this fix

Related Commands

MCP (Model Context Protocol) Integration

Connect to external tools and services through MCP servers. Access Slack, databases, APIs, and more.

MCP Integration

Usage Examples

Check MCP server configuration

Send a message to Slack via MCP

Related Commands

Browser Automation

Automate browser actions through Playwright MCP for web page verification and E2E testing.

MCP Integration

Usage Examples

Open localhost:3000 and take a screenshot

Run a form input test

Related Commands

External API Access

Access external web pages and APIs through Fetch MCP or custom MCP servers.

MCP Integration

Usage Examples

Fetch repository info from GitHub API

Read and summarize this URL's documentation

Database Access

Execute queries, check schemas, and manipulate data through MCP database servers.

MCP Integration

Usage Examples

Check the database schema

Get the latest 10 records from the users table

CLAUDE.md (Project Settings)

Configure project-specific instructions, coding conventions, and commands for Claude Code via the CLAUDE.md file.

Customization

Usage Examples

Create a CLAUDE.md for this project

Add coding conventions to CLAUDE.md

Related Commands

Skills (Custom Workflows)

Add custom workflows to Claude Code with SKILL.md files. Invoke with /skill-name or let Claude apply them automatically. Bundled skills like /batch and /simplify are available out of the box.

Customization

Usage Examples

/simplify to review and fix recent changes

/batch migrate src/ from Solid to React

Related Commands

Hooks (Event Handlers)

Attach shell commands, HTTP endpoints, LLM prompts, or agents to 20+ lifecycle events (PreToolUse, Stop, Notification, etc.). Automate pre-commit checks, notifications, and more.

Customization

Usage Examples

Set up a PostToolUse hook to auto-run lint after file edits

Add a Notification hook to send desktop alerts when Claude is idle

Related Commands

Permission Modes

Control tool execution permission levels. Adjust autonomy with plan, autoEdit, fullAuto, and more.

Customization

Usage Examples

Switch to plan mode to design the implementation

Switch to auto-execution mode

Related Commands

Model Selection

Switch between AI models (Opus/Sonnet/Haiku) based on the task. Useful for cost optimization.

Customization

Usage Examples

Switch to the Sonnet model

Haiku is sufficient for this task

Related Commands

Memory

Two complementary systems: CLAUDE.md files for manual instructions and Auto Memory for automatic learning across sessions. Claude remembers your corrections and preferences without manual effort.

Customization

Usage Examples

Remember to write commit messages in English

Remember that this project uses bun

Related Commands

Plugin System

Package skills, agents, hooks, and MCP servers as plugins to share with teams and the community. Install from marketplaces or local directories.

Customization

Usage Examples

claude plugin install code-review@claude-plugins-official

/reload-plugins to reload plugins

Related Commands

Keybinding Customization

Customize keyboard shortcuts via ~/.claude/keybindings.json. Rebind keys, add chord shortcuts, and configure vim mode bindings.

Customization

Usage Examples

Rebind Ctrl+S to submit

Check vim mode keybindings

Related Commands

Concise Output Style

A built-in "Concise" output style: Claude leads with the result and skips preamble and narration, while doing the work just as thoroughly. Pick it under Output style in `/config`, or set outputStyle in settings.json. Added in v2.1.237.

Customization

Usage Examples

Select Concise under Output style in /config

Set outputStyle to Concise in settings.json

Related Commands

Sub-Agents

Delegate tasks to specialized AI sub-agents. Built-in agents include Explore (search-optimized), Plan (read-only research), and general-purpose. Create custom sub-agents with .claude/agents/ files. From v2.1.232, fork mode is on by default in interactive sessions: a fork sub-agent inherits the parent's full conversation and prompt cache (off by default with -p and in the Agent SDK; toggle with CLAUDE_CODE_FORK_SUBAGENT).

Advanced

Usage Examples

Use an Explore agent to search the codebase

Create a custom agent in .claude/agents/

Related Commands

Parallel Agents

Run multiple sub-agents in parallel to handle independent tasks simultaneously. Ideal for concurrent research and implementation.

Advanced

Usage Examples

Run security review and code review in parallel

Create three translation files simultaneously

Agent Teams

Coordinate multiple Claude Code instances working as a team. A lead assigns tasks and teammates work independently with direct messaging between members.

Advanced

Usage Examples

Create an agent team to review PR #142 from security, performance, and test perspectives

Spawn 4 teammates to refactor modules in parallel

Context Management

Manage conversation history compression, efficient context window usage, and topic switching.

Advanced

Usage Examples

Compact the context

Clear to start a new topic

Related Commands

Checkpointing & Rewind

Automatically track file edits and rewind code and conversation to any previous point. Press Esc twice or use /rewind to access the checkpoint menu.

Advanced

Usage Examples

Press Esc Esc to open the rewind menu

/rewind to restore a previous state

Related Commands

IDE Integration

Use Claude Code within VS Code, JetBrains, and other IDEs via extensions and plugins.

Advanced

Usage Examples

Use as a VS Code extension

Use with a JetBrains plugin

Headless Mode

Execute tasks directly from the CLI without interaction. Ideal for CI/CD pipelines and script automation.

Advanced

Usage Examples

claude -p 'Write tests for this file' --output-file test.ts

echo 'Review this' | claude -p --json

Related Commands

SDK / API Integration

Call Claude Code features programmatically using the Claude Code SDK. Build custom tools and workflows.

Advanced

Usage Examples

Create a script that calls Claude Code via the TypeScript SDK

Build a custom agent

GitHub Actions Integration

Automate PR reviews and code generation using Claude Code in GitHub Actions.

Advanced

Usage Examples

Set up a workflow for auto-reviewing PRs

Create an Action that auto-generates PRs from issues

Claude Code on the Web

Run Claude Code from claude.ai/code in your browser. No local setup required — clone GitHub repos, make changes, and create PRs. Supports parallel task execution and diff review.

Advanced

Usage Examples

claude --remote 'Fix the login bug'

Teleport a web session to your terminal: claude --teleport

Related Commands

Remote Control

Control a local Claude Code session from a web browser or smartphone. Your filesystem and MCP servers stay local while you work from anywhere.

Advanced

Usage Examples

claude remote-control --name 'My Project'

/remote-control to make the session available remotely

Related Commands

Channels (Messaging Integration)

Push events from Telegram, Discord, iMessage, and custom webhooks into a running Claude Code session. Two-way communication for chat bridge workflows.

Advanced

Usage Examples

claude --channels plugin:telegram@claude-plugins-official

Send messages from Discord to Claude Code

Related Commands

Git Worktree Isolation

Give each session its own isolated copy of the repository using Git worktrees. Prevents file conflicts during parallel work. Sub-agents can also use worktree isolation.

Advanced

Usage Examples

claude --worktree feature-auth

Run agents in isolated worktrees for parallel work

Related Commands

Scheduled Tasks

Run Claude Code tasks on a recurring schedule. Four options: cloud (claude.ai), desktop app, GitHub Actions, and /loop command for in-session polling.

Advanced

Usage Examples

/loop 5m check if the deploy finished

Schedule a daily open PR review

Related Commands

Voice Dictation (Push-to-Talk)

Hold Space to dictate prompts with real-time transcription. Mix voice and typing freely. Supports 20 languages with coding-optimized recognition.

Advanced

Usage Examples

/voice to enable voice dictation

Hold Space to record, release to finalize

Related Commands

Fast Mode

Run Opus with faster output at the same model quality (as of v2.1.219, /fast applies to Opus 5 and Opus 4.8; Opus 4.7 was removed from fast mode). Higher cost per token but significantly lower latency. Ideal for interactive iteration and live debugging.

Advanced

Usage Examples

/fast to toggle Fast Mode

Enable fast mode for a debugging session

Related Commands

Fullscreen Rendering

Flicker-free alt-screen renderer for long sessions. Toggle with /tui fullscreen. Enables Ctrl+[ scrollback dump and v editor open in the transcript viewer. Works with /focus for a minimal view.

Advanced

Usage Examples

/tui fullscreen to relaunch into the alt-screen renderer

/focus to show only the last prompt and final response

Related Commands

Push Notifications

Claude can send mobile push notifications to your phone when Remote Control is enabled and 'Push when Claude decides' is configured. Stay informed about long-running work without watching the terminal.

Advanced

Usage Examples

Enable in /config → Push Notifications, then run with Remote Control

Claude notifies you when a long build finishes or when awaiting input

Related Commands

Goal-driven work

Set a completion condition with /goal and Claude keeps working across turns until the goal is met. Works in interactive mode, -p, and Remote Control, with a live overlay showing elapsed time, turns, and tokens. Added in v2.1.139.

Advanced

Usage Examples

/goal all tests pass — keep working until every test passes

/goal clear — remove the active goal early

Related Commands

Dynamic workflows

Ask Claude to create a workflow and it orchestrates tens to hundreds of agents in the background, so you can take on larger, more complex tasks that one context can't hold. Run /workflows to watch, pause, resume, or save runs. Added in v2.1.154. The trigger keyword was renamed from workflow to ultracode in v2.1.160 (asking in your own words still works). A "Dynamic workflow size" setting was added to /config in v2.1.202, giving Claude advisory guidance on how many agents to spawn (not a hard cap). In v2.1.219 that guidance now defaults to medium (aim for fewer than 15 agents); pick another size or unrestricted with "Dynamic workflow size" in /config.

Advanced

Usage Examples

"Create a workflow to migrate src/ from Solid to React" — decomposes the work and runs it in parallel

/workflows — monitor the progress of running and completed workflows

Related Commands

Claude in Chrome

Control a browser directly from the CLI through the Claude in Chrome extension. Your browser's login state is shared, so Claude can access sites you're already signed into. Chain browser actions with coding tasks in one workflow: read console errors, verify a UI against a design, or test form validation. Reached general availability in v2.1.198.

Advanced

Usage Examples

"Open localhost:3000, submit the form with invalid data, and check if the error messages appear"

"Open the dashboard page and check the console for errors when it loads"

Related Commands

Screen reader mode

An opt-in display mode for screen reader users that renders flat plain text without decorative borders or animations. Enable it with claude --ax-screen-reader, the CLAUDE_AX_SCREEN_READER=1 environment variable, or the axScreenReader: true setting. Added in v2.1.208.

Advanced

Usage Examples

claude --ax-screen-reader — start with flat plain-text rendering

Related Commands

Self-hosted execution environments

Run cloud sessions on your own infrastructure. claude self-hosted-runner starts one of your machines or containers as a runner, and Claude Code web, mobile, and desktop sessions then run on it. claude -p "..." --environment ccpool_... dispatches a session to that environment from a script. Public beta on Team and Enterprise plans. Added in v2.1.224.

Advanced

Usage Examples

claude self-hosted-runner --capacity 4 — start your machine as a runner that accepts up to 4 concurrent sessions

claude -p "Run the smoke test" --environment ccpool_abc123 — create a cloud session on that environment

Related Commands

Cross-session messaging

Running Claude Code sessions can message each other. SendMessage reaches a session on any of your machines, and ListAgents discovers the sessions you can send to. The crossSessionInbound setting holds messages sent to a session running with bypassed permissions for your approval. From v2.1.232, typing at least one letter after @ suggests your other live sessions on this machine, so a mention like @api-worker names the target directly, and the /config row "Messages from your other sessions" sets the inbound policy. macOS and Linux. Added in v2.1.224.

Advanced

Usage Examples

"Use ListAgents to see which sessions are running, then tell the API session about the spec change"

"SendMessage the test session on my other machine that the fix has landed"

Spellcheck

An opt-in setting that underlines misspelled words in the prompt input as you type, using whichever of aspell, hunspell, or ispell is installed. Turn it on with the spellcheck setting in settings.json (off by default). Added in v2.1.235.

Advanced

Usage Examples

Add the spellcheck setting to settings.json to enable it

Install a spell checker first, e.g. brew install aspell

Update History

A timeline of feature additions and changes by Claude Code version.

v2.1.261 (September 4, 2026)

Added /skill-doctor to show what each skill costs in context and how often it is used, so you can find skills to turn off, an extension of the existing skills feature rather than a standalone feature
Added --append-subagent-system-prompt-file to read the subagent system prompt append text from a file, for prompts too large to pass on the command line, an extension of the existing subagent features rather than a standalone feature
Added bashOutputMaxChars and taskOutputMaxChars settings to raise how much command and background-task output Claude receives inline before it is saved to a file, up to 128K characters, a setting on the existing settings feature rather than a standalone feature
Added an "Organization policy" line to /status and claude doctor that says why your organization's policy could not be loaded, such as a proxy not passing the endpoint through
No features were added or removed (the 71 features in this guide are unchanged)

v2.1.260 (September 3, 2026)

Added a diff panel that opens beside the conversation in fullscreen mode and shows your uncommitted changes as Claude edits; toggle it with /diff
Added a likely cause for prompt-cache misses (e.g. tool definitions or system prompt changed, idle past the TTL) to /cost and the status line's prompt_cache field
Added /reload-plugins to headless sessions, so it appears in the Claude Code Desktop and SDK command lists
Added a text form of /advisor (/advisor, /advisor <model>, /advisor off) for the desktop app, Remote Control, and other headless (-p/Agent SDK) sessions
Fixed Edit/Write/Read permission rules whose path contains parentheses being dropped as invalid or ignored by the Bash sandbox, which left "read-only" folders writable
Reverted the 2.1.259 change applying Read() deny rules to Bash arguments; it denied npm run build under a Read(./**/build/**) rule in every mode and made cd … && grep prompt even in auto mode
No features were added or removed (the 71 features in this guide are unchanged)

v2.1.259 (September 2, 2026)

Added the managedMcpServers managed setting so organizations can provide HTTP/SSE MCP servers to every user, a setting on the existing mcp-servers feature rather than a standalone feature
Added --permission-prompts none to deny all would-be prompts automatically on unattended headless hosts, while the active permission mode still decides everything else
Added recognition of glab mr create/merge/close/reopen/note/update so GitLab merge requests show as MR !N in the collapsed tool summary, strengthening the existing Git integration
Added --json to claude plugin validate for a machine-readable validation report
Fixed concurrent sessions silently reverting each other's ~/.claude.json changes, plus several worktree-isolation and background-workflow/agent stop-resume reliability issues
No features were added or removed (the 71 features in this guide are unchanged)

v2.1.258 (September 1, 2026)

Bug fixes only. Fixed a regression introduced in v2.1.255 that stopped Claude Code from launching on macOS 12 (Monterey)
Fixed remote and scheduled sessions failing after a re-sent permission approval could not be applied, improving remote-control reliability
No features were added or removed (the 71 features in this guide are unchanged; v2.1.253 through v2.1.256 were never released)

v2.1.257 (September 1, 2026)

Added Claude Fable 5.1 (claude-fable-5-1) as the new default Fable model with a 1M context, a model addition to the existing model selection rather than a standalone feature
Auto mode gained a Containment Escape rule and a one-time prompt before the first file read outside the working directories (permissions.blockReadsOutsideWorkingDirectories), strengthening the existing permission-mode feature
The new timeFormat and timeZone settings and the CLAUDE_CODE_SUBAGENT_MODEL_FORCE environment variable are options on existing settings management and subagent configuration, so they are not listed as standalone features
No features were added or removed (the 71 features in this guide are unchanged)

v2.1.252 (August 31, 2026)

Bug fixes only. Fixed Bash commands failing with "task output swap refused" on some Macs, and "always allow" not being saved in a project that has no .claude/settings.local.json yet
Fixed Remote Control sessions hosted by Claude Desktop or VS Code stalling for minutes after a tool finished when the connection to claude.ai was degraded, improving remote-control reliability
Fixed background task notifications with very large failure output pushing the conversation past the API request size limit
No features were added or removed (the 71 entries in this guide are unchanged)

v2.1.251 (August 28, 2026)

Added the PreModelSwitch and PostModelSwitch hook events, so a model switch can be blocked, confirmed, or annotated (new events on the existing hooks feature rather than a standalone feature)
Foreground subagent tool calls and their results now stream live to Remote Control clients, making remote-control sessions easier to follow as they run
Added a spend limit bar to /usage, and a per-session prompt cache line to /cost showing the cache hit rate and how many tokens were re-cached
Fixed many bugs, including security-related ones: a file tool permission bypass via symlinks, path traversal in plugins, and permission checks skipped by Bash arithmetic assignment
No features were added or removed (the 71 features in this guide are unchanged)

v2.1.250 (August 28, 2026)

Bug fixes and reliability improvements only; no new features (v2.1.249 was never released and is a gap in the sequence)
No features were added or removed (the 71 features in this guide are unchanged)

v2.1.248 (August 27, 2026)

Added --restricted (or CLAUDE_CODE_RESTRICTED=1), which starts Claude Code without the built-in tools that run commands or code and with file tools confined to the working directory (a startup option on the existing permission and tool controls, so it is not listed as a standalone feature; it is listed in the commands reference)
Cross-session messaging (SendMessage / ListAgents) between sessions on the same machine now works on Bedrock, Vertex, and Foundry, and when telemetry is disabled, widening where agent-teams applies
Added experimental.cacheTtl to agent frontmatter, an option on the existing subagent configuration rather than a standalone feature
The Workflow tool's prompt footprint dropped from about 5.7k tokens to about 1k, with the script-writing reference moved into a bundled workflow-authoring skill (an internal improvement to the workflows feature)
No features were added or removed (the 71 features in this guide are unchanged)

v2.1.247 (August 26, 2026)

Added the SendFeedback tool, which lets Claude draft feedback reports into the /feedback drafts queue (an extension of the existing feedback path, so it is not listed as a standalone feature)
/claude-api cost-optimize and the skill's new Admin API coverage are subcommand additions to a bundled skill, so they fall within the existing skills feature
Sonnet 5's default auto-compact window changed to its full 1M context, widening context-management behavior from about 934K to about 967K tokens
No features were added or removed (the 71 features in this guide are unchanged)

v2.1.246 (August 25, 2026)

/permissions gained an Auto mode tab for viewing and editing auto mode classifier rules, extending the existing permission-mode feature
/cd now applies the new directory's project settings, hooks, MCP servers, skills, and agents right after the move, so you no longer have to resume the session for them to take effect
No features were added or removed (still 71 features listed). v2.1.244 was never released, and v2.1.245 only fixed a startup crash on Linux with glibc 2.44

v2.1.243 (August 24, 2026)

/usage gained a Loops breakdown, so you can see token usage per /loop task
The new modelPicker, promptCacheTtl, and modelPricing settings extend existing model selection and settings management, so they are not listed as standalone features
No features were added or removed (still 71 features listed)

v2.1.241 (August 23, 2026)

Both v2.1.240 and v2.1.241 are bug-fix and reliability releases — no new features
No features were added or removed (the guide still covers 71 features)

v2.1.239 (August 21, 2026)

Added /claude-api upgrade, which migrates Python projects from anthropic 0.x to 1.x
In cloud sessions, plugins synced from claude.ai now show as name@synced and can be enabled or disabled individually
ListAgents now reports a session's own name and lists live teammates, making cross-session message targets easier to find
No features were added or removed (the guide still covers 71 features)

v2.1.238 (August 20, 2026)

Added a keybindingFlavor setting: set it to "readline" to make Ctrl+W in the prompt delete back to the previous whitespace, as in Bash (the default "classic" is unchanged)
Plugin marketplaces gained headersHelper, a command that mints HTTP headers for catalog and same-origin archive fetches
Ctrl+L and Cmd+K in fullscreen now always just repaint — the double-press /clear shortcut was removed
No features were added or removed (still 71 features)

v2.1.237 (August 20, 2026)

Added a built-in "Concise" output style: Claude leads with results and skips preamble and narration while doing the work just as thoroughly (feature guide: +1 entry — Concise output style)
Fixed prompt caching for sessions using an LLM gateway or a custom base URL

v2.1.236 (August 19, 2026)

Added the ANTHROPIC_DEFAULT_MODEL environment variable: it sets the model new sessions start on, while a /model pick still overrides it and persists across restarts (unlike ANTHROPIC_MODEL)
Added notify_when_idle to cross-session SendMessage: ask another Claude Code session on this machine to send one notice when it next goes idle (opt-in, one-shot, macOS and Linux)
Sandbox on macOS: wildcard read-deny rules such as **/.env now take precedence inside allowed read regions and can no longer be bypassed by renaming the denied file
[VSCode] Added screen reader support for the transcript: live announcements for replies, permission requests, errors, and status changes, plus per-turn heading navigation
No features were added, removed, or recategorized in this guide

v2.1.235 (August 18, 2026)

Added an optional "spellcheck" setting that underlines misspelled words in the prompt input as you type (feature guide: +1 entry — spellcheck)
Fixed the Agent tool advertising a general-purpose default in sessions where that agent is unavailable
Improved memory and CPU usage for background cloud sessions such as /ultrareview

v2.1.234 (August 17, 2026)

A session that hit a claude.ai usage limit now continues automatically when the limit resets; turn it off with "Continue automatically at usage limit" in /config
GitLab integration reaches further: repos with a GitLab remote and an authenticated glab CLI now show a merge request badge (MR !N) with draft, pending, and green states in the footer and statusline
The built-in claude-api skill now costs about 25k tokens to load instead of 200k+, because its reference docs load on demand
Added the CLAUDE_CODE_PROJECT_DIR_NAME environment variable so hosts that give each session its own config directory can pick a short name for the per-project transcript directory
Improved the transcript: your own prompts now render markdown (highlighted code blocks, inline code, lists) the same way replies do
No features were added, removed, or recategorized in this guide

v2.1.233 (August 14, 2026)

Todo and task-management tools are now off by default on Opus 4.8, Sonnet 5, Fable 5, Mythos 5, and later models. Set CLAUDE_CODE_ENABLE_TODO_TOOLS=1 to keep them enabled
Git integration reaches further: --worktree and the claude agents view now accept GitLab merge request URLs (creating a branch from a GitLab merge request requires v2.1.233 or later)
Two new environment variables: CLAUDE_CODE_TOOL_MEMORY_LIMIT caps Bash tool memory through Linux cgroups, and CLAUDE_CODE_WEBFETCH_CACHE_TTL_MS sets the TTL for the per-session URL cache used by WebFetch. Both are opt-in, so default behavior is unchanged
The /effort selector in screen reader mode now renders as a numbered list with a number-entry prompt
No features were added, removed, or recategorized in this guide

v2.1.232 (August 13, 2026)

Subagent forking is now on by default: a fork sub-agent inherits the parent's full conversation and prompt cache, making it cheaper than a fresh sub-agent for tasks that need the same context (interactive sessions only — still off with -p and in the Agent SDK)
In interactive sessions, non-teammate agent spawns now run in the background by default
Cross-session messaging gained @ mentions: type at least one letter after @ and your other live sessions on this machine appear as suggestions, so @api-worker names the target
Interactive sessions on one machine now keep unique names — reusing a live session's name assigns a name-word-word variant instead
Plugin marketplaces now support GitLab: bare gitlab.com repo URLs, including nested subgroups, clone like github.com URLs
No features were added, removed, or recategorized in this guide

v2.1.231 (August 13, 2026)

Fixed MCP OAuth sign-in failing with a redirect URI mismatch for servers that use a pre-registered OAuth client, such as Slack
No new features, and no changes to the entries in this guide

v2.1.229 (August 12, 2026)

No new features. This release focuses on bug fixes and internal improvements
Confirmed --exec while re-checking the official CLI reference: combined with --bg, it runs a shell command as a PTY-backed background job instead of starting a Claude session (listed in the commands reference, since it is a CLI flag)
claude remote-control --continue is now documented in the official docs
No features were added, removed, or recategorized in this guide

v2.1.228 (August 11, 2026)

No new features. This release focuses on bug fixes and security hardening
Hardened skills security: a skill synced from claude.ai can no longer override a local command or run shell commands
Fixed interactive sessions freezing mid-render and Git Bash detection on Windows
Fixed /tui not restoring the model after switching with /model, and cross-session messages missing from the inbox
Fixed cleanup bugs in memory management and the plugin cache

v2.1.227 (August 10, 2026)

No new features. This release focuses on bug fixes and improvements to existing capabilities
Fixed feature flags being evaluated before the subscription tier was resolved when a session started with an expired login token
Fixed /tui restoring conversations that had been rewound to before their first message
Improved the slash command menu UI: only the selected row is highlighted, matched characters are bolded, and glyphs in emoji and accented names are preserved
Reduced event loop stalls in file-not-found suggestions and @-mention size checks

v2.1.225-v2.1.226 (August 2026)

No new features. These releases focus on bug fixes and improvements to existing capabilities
Added a workspace trust prompt to claude agents for untrusted directories, matching the behavior of claude (v2.1.225)
Cross-session SendMessage can now start a conversation with your Remote Control sessions on other machines by name, instead of only replying after they message you first (covered by the existing cross-session messaging entry; v2.1.225)
Fixed self-hosted environments registering and then failing every session when --base-dir cannot be created or written (v2.1.225)

v2.1.224 (August 2026)

Added self-hosted environments: claude self-hosted-runner turns your own machines or containers into a place Claude Code web, mobile, and desktop sessions can run (Team and Enterprise plans)
Added cross-session SendMessage: sessions can now message each other on any of your machines, with ListAgents to discover them (macOS and Linux)
Added an archive plugin source: install plugins from a zip over HTTPS without git or npm, with optional SHA-256 pinning
Removed the 200-subagent-per-session spawn cap (concurrency and depth limits still apply)

v2.1.223 (August 2026)

A bug-fix and security-hardening release with no new features
Security fixes: crafted commands could hide part of a Bash permission check; tabs and invisible Unicode could hide part of a command in the permission prompt; workflow scripts could run code outside the sandbox via dynamic import()
Changed /review to be an alias of /code-review (already covered by the existing code review capability entry)

v2.1.222 (August 2026)

Removed the ultraplan feature (use plan mode instead)
Fixed worktree-isolated sessions and their subagents being able to run destructive git commands against the main checkout; isolation now applies to file edits and Bash in every session type
Fixed PreToolUse auto-allow hooks bypassing tool restrictions in background agent tasks (summaries, compaction, renames)
Changed Remote Control auto-start so repo-local settings (.claude/settings.json or .claude/settings.local.json) can no longer turn it on (they can still turn it off); enable it at user scope via /config

v2.1.221 (August 2026)

Added mode: "mask" for sandbox credential files on Linux and WSL: sandboxed commands read a sentinel copy (the whole file, or just the spans captured by an extract regex) while the sandbox proxy substitutes the real value on egress; on macOS, file masking falls back to deny
Added a prompt-audit subcommand to the claude-api skill for auditing prompts and tool descriptions for patterns written for older models
Improved cache efficiency in auto-mode permission checks by reusing the cached conversation prefix across decisions, reducing prompt-cache costs

v2.1.219-v2.1.220 (July 2026)

Added Claude Opus 5 (claude-opus-5), now the default Opus model — 1M context, with fast mode available (v2.1.219)
Changed subagents to spawn nested subagents up to depth 3 by default (was 1, i.e. nesting off); set CLAUDE_CODE_MAX_SUBAGENT_SPAWN_DEPTH=1 to disable nesting (v2.1.219)
Changed dynamic workflows to default to a medium size guideline (aim for fewer than 15 agents); pick another size or unrestricted with "Dynamic workflow size" in /config (v2.1.219)
Added nested subagent forwarding in stream-json: subagents spawned at depth-2+ now appear when --forward-subagent-text is set, keyed by their spawning Agent tool_use id (v2.1.219)

v2.1.218 (July 2026)

Changed /code-review to run as a background subagent, so review work no longer fills your conversation (v2.1.218)
Added screen-reader announcements of deleted text for word and line deletions in --ax-screen-reader mode (v2.1.218)
Added HTTP status and error text to claude mcp list and /mcp when a server fails to connect (v2.1.218)
Changed skills with context: fork to run in the background by default; opt out per skill with background: false (v2.1.218)

v2.1.217 (July 2026)

Added a cap on concurrently-running subagents (default 20, override with CLAUDE_CODE_MAX_CONCURRENT_SUBAGENTS) so one message can't fan out unbounded background agents (v2.1.217)
Changed subagents to no longer spawn nested subagents by default; set CLAUDE_CODE_MAX_SUBAGENT_SPAWN_DEPTH to allow deeper nesting (v2.1.217)
Added emoji shortcode autocomplete in the prompt input (:heart: inserts ❤️, disable with emojiCompletionEnabled) (v2.1.217)

v2.1.216 (July 2026)

Added the sandbox.filesystem.disabled setting to skip filesystem isolation while keeping network egress control (v2.1.216)
Fixed a slowdown in long sessions where message normalization cost grew quadratically with turns (v2.1.216)

v2.1.214-v2.1.215 (July 2026)

Added the EndConversation tool so Claude can end sessions with highly abusive users or jailbreak attempts (v2.1.214)
Hardened the permission system: added prompts for docker commands with daemon-redirect flags, and stopped Bash permission checks from auto-approving commands over 10,000 characters, file-descriptor redirect forms, and similar edge cases (v2.1.214)

v2.1.211-v2.1.212 (July 2026)

Changed /fork to copy the conversation into a new background session instead of launching an in-session subagent; the in-session role moved to the new /subtask command (v2.1.212)
Added session-wide safety limits on WebSearch calls and subagent spawns to stop runaway loops, and MCP tool calls over 2 minutes now move to the background automatically (v2.1.212)

v2.1.209-v2.1.210 (July 2026)

Added a live elapsed-time counter to the collapsed tool summary line so long-running tool calls visibly progress (v2.1.210)
Hardened the Agent tool against indirect prompt injection via content a subagent reads (v2.1.210)

v2.1.208 (July 2026)

Added screen reader mode: opt-in plain-text rendering for screen reader users (v2.1.208)
Reduced per-tool-call CPU overhead in print/SDK sessions with many MCP tools, up to 7x faster tool rounds (v2.1.208)

v2.1.207 (July 2026)

Auto mode is now available without CLAUDE_CODE_ENABLE_AUTO_MODE opt-in on Bedrock, Vertex AI, and Foundry (v2.1.207)
Changed Bedrock, Vertex, and Claude Platform on AWS to default to Claude Opus 4.8 (v2.1.207)

v2.1.205 (July 2026)

/doctor is now a comprehensive setup checkup, with /checkup available as an alias (v2.1.205)
Improved auto mode to ask before running rm -rf on unresolved variables (v2.1.205)

v2.1.203 (July 2026)

Added the session's additional working directories to MCP roots/list, with notifications/roots/list_changed sent when the set changes (v2.1.203)
Added a grey ⏸ badge to the footer when in manual permission mode, making the active permission mode always visible (v2.1.203)

v2.1.202 (July 2026)

Added a "Dynamic workflow size" setting to /config, giving Claude advisory guidance on how many agents a dynamic workflow should spawn (small/medium/large), not a hard cap (v2.1.202)

v2.1.198 (July 2026)

Claude in Chrome browser automation reached general availability (v2.1.198)
Added the /dataviz skill for chart and dashboard design guidance with a runnable color-palette validator (v2.1.198)
Background agents launched from claude agents now commit, push, and open a draft PR when they finish code work in a worktree, instead of stopping to ask (v2.1.198)

v2.1.197 (June 2026)

Introduced Claude Sonnet 5 as the new default model in Claude Code, with a native 1M-token context window and promotional pricing of $2/$10 per Mtok through August 31 (v2.1.197)

v2.1.196 (June 2026)

Added support for organization default models — admins set it in the org console, and it shows as "Org default" (or "Role default") in /model when you haven't picked one yourself (v2.1.196)
Made chat file attachments clickable — Cmd/Ctrl-click reveals the file in Finder/Explorer (v2.1.196)
Enabled the streaming idle watchdog by default for all providers — it aborts and retries when a response stream produces no events for 5 minutes (set CLAUDE_ENABLE_STREAM_WATCHDOG=0 to disable) (v2.1.196)

v2.1.195 (June 2026)

Added CLAUDE_CODE_DISABLE_MOUSE_CLICKS to disable mouse click/drag/hover in fullscreen mode while keeping wheel scroll (v2.1.195)
Fixed hook matchers with hyphenated identifiers (e.g. code-reviewer, mcp__brave-search) accidentally substring-matching — they now exact-match; use mcp__brave-search__.* to match all tools from a hyphenated MCP server (v2.1.195)
Improved the claude agents completed list to fill available vertical space and added a provisioning checklist to Remote session startup (v2.1.195)

v2.1.193 (June 2026)

Added autoMode.classifyAllShell setting to route all Bash/PowerShell commands through the auto-mode classifier instead of only arbitrary-code-execution patterns (v2.1.193)
Added live file path autocomplete to bash mode (!) (v2.1.193)
Added auto-mode denial reasons to the transcript, the denial toast, and /permissions recent denials (v2.1.193)

v2.1.191 (June 2026)

Added /rewind support for resuming a conversation from before /clear was run (v2.1.191)
Reduced CPU usage during streaming responses by ~37% by coalescing text updates to 100ms (v2.1.191)
Improved MCP server reliability: capability discovery (tools/list, prompts/list, resources/list) now retries transient network errors with short backoff (v2.1.191)

v2.1.187 (June 2026)

Added the sandbox.credentials setting to block sandboxed commands from reading credential files and secret environment variables (v2.1.187)
Added org-configured model restrictions to the model picker, --model, /model, and ANTHROPIC_MODEL, with a "restricted by your organization's settings" message when a restricted model is selected (v2.1.187)
Added mouse click support to select menus (permission prompts, /model, /config, etc.) in fullscreen mode (v2.1.187)

v2.1.186 (June 2026)

Added claude mcp login <name> / claude mcp logout <name> to authenticate and log out of MCP servers from the CLI without opening the /mcp menu (with --no-browser for SSH/headless sessions) (v2.1.186)
Changed background subagents to surface permission prompts in the main session instead of auto-denying; the dialog shows which agent is asking, and Esc denies just that tool (v2.1.186)
Changed ! bash commands to trigger Claude to respond to the output automatically; set "respondToBashCommands": false in settings.json to keep the previous context-only behavior (v2.1.186)

v2.1.183 (June 2026)

Improved auto mode safety: destructive git commands (git reset --hard, git checkout -- ., git clean -fd, git stash drop) are blocked unless you asked to discard local work, git commit --amend is blocked when the commit wasn't made by the agent this session, and terraform destroy/pulumi destroy/cdk destroy are blocked unless you asked for the specific stack (v2.1.183)
Added a warning when the requested model is deprecated or automatically updated, shown on stderr in print mode (-p) and covering models set in agent frontmatter (v2.1.183)
Added the attribution.sessionUrl setting to omit the claude.ai session link from commits and PRs in web and Remote Control sessions (v2.1.183)

v2.1.181 (June 2026)

Added /config key=value syntax to set any setting directly from the prompt (e.g. /config thinking=false; works in interactive, -p, and Remote Control) (v2.1.181)
Added the sandbox.allowAppleEvents opt-in setting that lets sandboxed commands send Apple Events on macOS (v2.1.181)
Added the CLAUDE_CLIENT_PRESENCE_FILE environment variable to suppress mobile push notifications while you're at the machine (v2.1.181)
Improved the subagent panel: idle subagents auto-hide after 30s and the list caps at 5 rows with scroll hints (v2.1.181)

v2.1.178 (June 2026)

Added Tool(param:value) syntax for permission rules to match a tool's input parameters (with * wildcard), e.g. Agent(model:opus) to block Opus subagents (v2.1.178)
Skills in nested .claude/skills directories now load when working on files there; on a name clash the nested skill appears as <dir>:<name> so both stay available (v2.1.178)
Improved auto mode: subagent spawns are now evaluated by the classifier before launch, closing a gap where a subagent could request a blocked action without review (v2.1.178)

v2.1.176 (June 2026)

Session titles are now generated in the language of your conversation (pin a specific language with the language setting) (v2.1.176)
Added the footerLinksRegexes setting for regex-matched link badges in the footer row (v2.1.176)

v2.1.175 (June 2026)

Added the enforceAvailableModels managed setting — the availableModels allowlist now also constrains the Default model (v2.1.175)

v2.1.174 (June 2026)

Added the wheelScrollAccelerationEnabled setting to disable mouse-wheel scroll acceleration in fullscreen mode (v2.1.174)
Fixed the /model picker hiding the model family that Default resolves to (v2.1.174)

v2.1.173 (June 2026)

Fixed Fable 5 model names with a [1m] suffix not being normalized — Fable 5 includes 1M context by default (v2.1.173)

v2.1.172 (June 2026)

Sub-agents can now spawn their own sub-agents, nested up to 5 levels deep (v2.1.172)
Added a search bar when browsing a marketplace's plugins in /plugin (v2.1.172)
Fixed 1M-context sessions without usage credits getting permanently stuck — they now automatically compact back under the standard context limit (v2.1.172)

v2.1.170 (June 2026)

Introduced Claude Fable 5, a Mythos-class model available via claude --model fable and --advisor fable, with capabilities exceeding any previously general-release model (v2.1.170)
Fixed sessions not saving transcripts (and missing from --resume) when launched from the VS Code integrated terminal or any shell that inherited Claude Code environment variables (v2.1.170)

v2.1.168-v2.1.169 (June 2026)

Added the --safe-mode flag (and CLAUDE_CODE_SAFE_MODE) to start with all customizations (CLAUDE.md, plugins, skills, hooks, MCP servers) disabled so you can isolate a broken configuration (v2.1.169)
Added the /cd command to move a session to a new working directory without breaking the prompt cache mid-session (v2.1.169)
Added a disableBundledSkills setting and CLAUDE_CODE_DISABLE_BUNDLED_SKILLS environment variable to hide bundled skills, workflows, and built-in slash commands from the model (v2.1.169)

v2.1.165-v2.1.167 (June 2026)

Added the fallbackModel setting to configure up to three fallback models tried in order when the primary model is overloaded or unavailable; --fallback-model now also applies to interactive sessions (v2.1.166)
Added glob pattern support in the deny rule tool-name position ("*" denies all tools) (v2.1.166)
MAX_THINKING_TOKENS=0, --thinking disabled, and the per-model thinking toggle now disable thinking on models that think by default via the Claude API (v2.1.166)

v2.1.162-v2.1.163 (June 2026)

Hooks: Stop and SubagentStop hooks can now return hookSpecificOutput.additionalContext to give Claude feedback and keep the turn going without being labeled a hook error (v2.1.163)
Skills: added \$ escape syntax to include a literal $ before a digit in command bodies (v2.1.163)
Added requiredMinimumVersion and requiredMaximumVersion managed settings — Claude Code refuses to start if its version is outside the allowed range and directs the user to an approved version (v2.1.163)
WebFetch permission rules now apply to built-in preapproved domains; explicit WebFetch(domain:...) deny/ask/allow rules now take precedence over the preapproved-host auto-allow (v2.1.162)

v2.1.160-v2.1.161 (June 2026)

Renamed the dynamic-workflow trigger keyword from workflow to ultracode — the word "workflow" no longer triggers a run (v2.1.160)
Security hardening: prompts before writing to shell startup files (.zshenv, etc.) and build-tool config files that grant code execution (.npmrc, etc.) (v2.1.160)
Single-file grep commands now satisfy the read-before-edit check, so Edit no longer requires a separate Read (v2.1.160)
Parallel tool calls: a failed Bash command no longer cancels other calls in the same batch (v2.1.161)
/mcp now collapses claude.ai connectors you've never signed in to behind a "Show unused connectors" row (v2.1.161)
OTEL_RESOURCE_ATTRIBUTES values are now included as labels on metric datapoints, so you can slice usage metrics by team or repo (v2.1.161)

v2.1.157-v2.1.158 (May 2026)

Plugins in .claude/skills directories now load automatically, no marketplace required (v2.1.157)
Added claude plugin init <name> — scaffolds a new plugin at ~/.claude/skills/<name>/ that auto-loads the next session (v2.1.157)
claude agents now honors the agent field in settings.json for dispatched sessions, with --agent <name> to override it (v2.1.157)
EnterWorktree can now switch between Claude-managed worktrees mid-session (v2.1.157)
Auto mode is now available on Bedrock, Vertex, and Foundry for Opus 4.7 and 4.8 — opt in with CLAUDE_CODE_ENABLE_AUTO_MODE=1 (v2.1.158)

v2.1.153-v2.1.156 (May 2026)

Introduced dynamic workflows — ask Claude to create a workflow and it orchestrates tens to hundreds of agents in the background. Run /workflows to watch, pause, resume, or save runs (v2.1.154)
Added claude --exec (with --bg) to run a shell command as a PTY-backed background job with attach/detach support (v2.1.154)
/simplify is now a cleanup-only review — reviews reuse, simplification, efficiency, and altitude in parallel and applies fixes, without hunting for bugs (v2.1.154)
Added the ultracode level to /effort, combining xhigh reasoning with automatic workflow orchestration (v2.1.154)
Added Opus 4.8 — defaults to high effort, with far cheaper fast mode (v2.1.154)
/model now saves your selection as the default; press s in the picker to switch for the current session only (v2.1.153)

v2.1.146-v2.1.152 (May 2026)

Renamed /simplify to /code-review — reports correctness bugs at a chosen effort level; pass --comment for inline GitHub PR comments and --fix to apply improvements to the working tree (v2.1.146-152)
Added /reload-skills to re-scan skill directories without restarting the session (v2.1.152)
Skills and slash commands can set disallowed-tools in frontmatter; SessionStart hooks can return reloadSkills and sessionTitle (v2.1.152)
Added the MessageDisplay hook event to transform or hide assistant message text as it is displayed (v2.1.152)
/usage now shows a per-category breakdown (skills, subagents, plugins, MCP servers) with streaming reads for large session files (v2.1.149/152)
/diff detail view supports keyboard scrolling. Markdown output renders GFM task list checkboxes (v2.1.149)
Switches to --fallback-model for the rest of the session when the primary model is not found, and Auto mode no longer requires opt-in consent (v2.1.152)

v2.1.143-v2.1.145 (May 2026)

Added run-and-verify capabilities — /run (launch and drive the app to see a change working), /verify (build and run to confirm a change works), and /run-skill-generator (write a per-project skill), three bundled skills (v2.1.145)
Added claude agents --json to list live Claude sessions as a JSON array for scripting (v2.1.145)
/plugin Discover and Browse screens now show a plugin's commands, agents, skills, hooks, and MCP/LSP servers before installation (v2.1.145)
/resume now supports background sessions and /model changes the model for the current session only; /extra-usage renamed to /usage-credits (v2.1.144)
Added plugin dependency enforcement, the worktree.bgIsolation: "none" setting, and auto mode in the Shift+Tab cycle for attached agent sessions (v2.1.143)

v2.1.142 (May 2026)

Added new claude agents flags (--add-dir, --settings, --mcp-config, --plugin-dir, --permission-mode, --model, --effort, --dangerously-skip-permissions) to configure dispatched background sessions
Fast mode now uses Opus 4.7 by default (previously Opus 4.6); set CLAUDE_CODE_OPUS_4_6_FAST_MODE_OVERRIDE=1 to pin it to Opus 4.6
Plugins with a root-level SKILL.md and no skills/ subdirectory are now surfaced as a skill
The /plugin details pane and claude plugin details now show the LSP servers a plugin provides
/web-setup warns before replacing an existing GitHub App connection
Improved reactive compaction — the first summarize attempt seeds from the original request's overflow size, avoiding a wasted near-full-context retry
Fixed MCP_TOOL_TIMEOUT not raising the timeout for remote HTTP/SSE MCP servers (capped at 60s), background sessions disappearing after macOS sleep/wake, and the daemon not exiting cleanly after a binary upgrade

v2.1.141 (May 2026)

Added terminalSequence field to hook JSON output so hooks can emit desktop notifications, window titles, and bells without a controlling terminal
Added CLAUDE_CODE_PLUGIN_PREFER_HTTPS environment variable to clone GitHub plugin sources over HTTPS for environments without an SSH key
Added ANTHROPIC_WORKSPACE_ID environment variable for workload identity federation — scopes the minted token to a specific workspace
Added claude agents --cwd <path> to scope the session list to a directory
/feedback can now include recent sessions (last 24 hours or 7 days) for issues spanning more than the current session
Rewind menu: added "Summarize up to here" to compress earlier context while keeping recent turns intact
Auto mode permission dialog now explains when a permissions.ask rule caused the prompt
Background agents launched via /bg or ←← now preserve the current permission mode instead of reverting to default
Improved spinner feedback during long thinking — spinner warms to amber after 10 seconds to signal Claude is still working
Improved plugin menu navigation (→/Tab switch tabs, ↑ to tab strip, clickable tab headers and search box in fullscreen)
Fixed /tui silently dropping running background shells and subagents — now refuses and asks to wait for them to finish
Fixed /model in one session silently changing the autocompact threshold in other sessions, markdown tables with cell wrapping falling back to vertical layout, Ctrl+C not interrupting in vim INSERT/VISUAL mode, /mcp focus retention, MCP HTTP/SSE 403 display, and Windows Alt+V image paste issues

v2.1.140 (May 2026)

Agent tool subagent_type matching is now case- and separator-insensitive ("Code Reviewer" resolves to code-reviewer)
Updated agent color palette
Fixed /goal silently hanging when disableAllHooks or allowManagedHooksOnly is set (now shows a clear message)
Fixed /loop scheduling redundant wakeups for background tasks that already notify on completion
Plugins warn in /doctor, claude plugin list, and /plugin when a default component folder (e.g. commands/) is silently ignored because plugin.json sets the matching key
Fixed claude --bg failing with "connection dropped mid-request" near background-service idle-exit, and startup failures on enterprise endpoint security
Fixed remote managed settings not retrying on 401 (now retries once with a force-refreshed token)
Fixed settings hot-reload misattributing change events for symlinked files, and Windows event-loop stalls from synchronous where.exe re-spawns
Fixed Read tool validation for whitespace-padded or +-prefixed offset strings, and the native terminal cursor not staying at the input caret on focus loss

v2.1.139 (May 2026)

Added /goal [condition|clear] — set a completion condition and Claude keeps working across turns until it's met. Shows live elapsed time, turns, and tokens as an overlay
Added /scroll-speed — tune mouse wheel scroll speed with a live preview ruler (fullscreen only)
Added transcript-view navigation keys: ? for the shortcut help panel, { / } to jump between user prompts, v to open the conversation in $VISUAL / $EDITOR
Added agent view as a Research Preview: claude agents lists every Claude Code session — running, blocked on you, or done
Added claude plugin details <name> to show a plugin's component inventory and projected per-session token cost
Added hook args: string[] (exec form, no shell, no quoting) and continueOnBlock (feed hook's rejection back to Claude on PostToolUse)
MCP stdio servers now receive CLAUDE_PROJECT_DIR in their environment
/mcp Reconnect picks up .mcp.json edits without a restart and shows HTTP status/URL on failure
Remote Control, /schedule, claude.ai MCP connectors, and notifications are disabled when ANTHROPIC_API_KEY / apiKeyHelper / ANTHROPIC_AUTH_TOKEN is set
Many fixes across hot-reload, OAuth, permissions, mouse wheel scrolling, border-embedded text, Bash history, and heap usage

v2.1.138 (May 2026)

Added CLAUDE_CODE_ENABLE_FEEDBACK_SURVEY_FOR_OTEL to re-enable the session quality survey for enterprises capturing responses through OpenTelemetry
Added settings.autoMode.hard_deny for auto mode classifier rules that block unconditionally regardless of user intent or allow exceptions
Fixed MCP servers configured in .mcp.json, plugins, and claude.ai connectors silently disappearing after /clear in the VS Code extension, JetBrains plugin, and Agent SDK
Fixed a rare login loop where a concurrent credential write could overwrite a freshly-rotated OAuth token and force re-login
Fixed MCP OAuth refresh tokens being lost when multiple servers refresh concurrently — users with several remote MCP servers should no longer need daily re-authentication
Fixed an API error (400) when extended thinking emitted a redacted thinking block after a tool call
Fixed --resume / --continue not finding sessions when the project path contains underscores, and plan mode not blocking file writes when a matching Edit(...) allow rule exists
WSL2: image paste from Windows clipboard now works via a PowerShell fallback when xclip/wl-paste cannot read image data
Improved visual consistency across slash command dialogs (footer hints, dialog spacing, arrow-key styling, immediate frame display during loading)
Fixed many UI/UX issues including /usage weekly reset display, CJK welcome banner overflow, /insights crash with malformed input fields, /branch saving multi-line session titles, AskUserQuestion discarding multi-select array answers, /clear <name> labeling, CronList output qualifiers, plugin slash command namespace resolution, Bash permission prompt parser diagnostics, /release-notes getting stuck on old versions, and /mcp list scrolling
VSCode extension Windows activation fix (v2.1.137) and internal fixes (v2.1.138)

v2.1.133 (May 2026)

Added worktree.baseRef setting (fresh | head) to choose whether --worktree, EnterWorktree, and agent-isolation worktrees branch from origin/<default> or local HEAD
The default fresh changes EnterWorktree's base back to origin/<default> (it has been local HEAD since 2.1.128) — set worktree.baseRef: "head" to keep unpushed commits in new worktrees
Added sandbox.bwrapPath and sandbox.socatPath managed settings (Linux/WSL) to specify custom bubblewrap and socat binary locations
Added parentSettingsBehavior admin-tier key ('first-wins' | 'merge') to let admins opt SDK managedSettings (parent tier) into the policy merge
Hooks now receive the active effort level via the effort.level JSON input field and the $CLAUDE_EFFORT environment variable, and Bash tool commands can read $CLAUDE_EFFORT
Improved focus mode behavior
Improved memory usage by releasing warm-spare background workers under memory pressure
Fixed parallel sessions all dead-ending at 401 after a refresh-token race wiped shared credentials
Fixed Edit/Write allow rules scoped to a drive root (C:\) or POSIX / matching incorrectly and always prompting
Fixed an unhandled rejection (ECOMPROMISED) when a history or session-log file lock is compromised by clock skew or slow disk
Fixed pressing Esc during conversation compaction showing a spurious "Error compacting conversation" notification
Fixed HTTP(S)_PROXY / NO_PROXY / mTLS not being respected for the full MCP OAuth flow including discovery, dynamic client registration, token exchange, and token refresh
Fixed Read/Write/Edit being denied on mapped network drives passed via --add-dir / SDK additionalDirectories
Fixed Remote Control stop/interrupt from claude.ai not fully canceling the CLI session the same way local Esc does
Fixed /effort in one session unexpectedly changing the effort level of other concurrent sessions, and a related issue where an IDE effort change could be silently dropped
Fixed subagents not discovering project, user, or plugin skills via the Skill tool
claude --help now lists --remote-control alongside --remote-control-session-name-prefix
[VSCode] Fixed claudeCode.claudeProcessWrapper failing with "Unsupported platform" when the extension build doesn't bundle a Claude binary

v2.1.132 (May 2026)

Added CLAUDE_CODE_SESSION_ID env var to the Bash tool subprocess environment, matching the session_id passed to hooks
Added CLAUDE_CODE_DISABLE_ALTERNATE_SCREEN=1 env var to opt out of the fullscreen alternate-screen renderer and keep the conversation in the terminal's native scrollback
Added a "Pasting…" footer hint while a Ctrl+V image paste is being read from the clipboard
Fixed external SIGINT (e.g. IDE stop button, kill -INT) not running graceful shutdown — terminal modes are now restored and the --resume hint is printed
Fixed --resume failing with no low surrogate in string when a tool error truncation split an emoji; pre-corrupted sessions are sanitized on load
Fixed --permission-mode flag being ignored when resuming a plan-mode session with -p --continue/--resume, and plan mode not being re-applied after ExitPlanMode
Fixed fullscreen mode showing a blank screen after laptop sleep/wake or Ctrl+Z/fg, and cursor landing mid-grapheme on Ctrl+E/A/K/U/arrow keys when an Indic conjunct or ZWJ emoji wraps across lines
Fixed /usage Ctrl+S hanging on Linux/X11 when copying the stats screenshot, /effort picker not honoring CLAUDE_CODE_EFFORT_LEVEL, and /status showing the wrong default model
Fixed unbounded memory growth (10GB+ RSS) when stdio MCP servers wrote non-protocol data to stdout, and MCP tools/list failures silently showing 0 tools
Fixed Bedrock and Vertex 400 errors when ENABLE_PROMPT_CACHING_1H is set
v2.1.131: Fixed VS Code extension failing to activate on Windows due to a bundled SDK createRequire polyfill bug, and Mantle endpoint authentication failing with missing x-api-key header

v2.1.129 (May 2026)

Added --plugin-url <url> flag to fetch a .zip plugin archive from a URL for the current session only — repeatable (--plugin-url A --plugin-url B)
Added CLAUDE_CODE_FORCE_SYNC_OUTPUT=1 env var to force-enable synchronized output on terminals where auto-detection misses (e.g. Emacs eat)
Added CLAUDE_CODE_PACKAGE_MANAGER_AUTO_UPDATE env var for Homebrew/WinGet installs — Claude Code runs the upgrade command in the background and prompts to restart
Plugin manifests: themes and monitors should now be declared under "experimental": { ... } — top-level declarations still work but claude plugin validate will warn
Gateway /v1/models discovery for the /model picker is now opt-in via CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1 (was automatic in 2.1.126–2.1.128)
Ctrl+R history picker now defaults to searching all prompts across all projects, matching pre-2.1.124 behavior — press Ctrl+S to narrow to the current project or session
skillOverrides setting now works: off hides from model and /, user-invocable-only hides from model only, name-only collapses description
The claude_code.pull_request.count OTel metric now counts PRs/MRs created via MCP tools, not just shell commands
Policy refusal error messages now include the API Request ID for easier support debugging
Many bug fixes including agent panel hidden during sub-agents (regression in 2.1.122), /context ASCII grid wasting ~1.6k tokens per call, 1-hour prompt cache TTL silently downgraded to 5 minutes, server-managed settings policy for users without user:inference scope, and OAuth refresh race after wake-from-sleep

v2.1.128 (May 2026)

Bare /color (no args) now picks a random session color
/mcp now shows the tool count for connected servers and flags servers that connected with 0 tools
--plugin-dir now accepts .zip plugin archives in addition to directories
--channels now works with console (API key) authentication — console orgs with managed settings must set channelsEnabled: true to enable
Updated /model picker: collapsed duplicate Opus 4.7 entries, and current Opus now shows as "Opus" instead of "Opus 4.7"
Subprocesses (Bash, hooks, MCP, LSP) no longer inherit OTEL_* environment variables, so OTEL-instrumented apps run via the Bash tool no longer pick up the CLI's own OTLP endpoint
MCP: workspace is now a reserved server name, and reconnecting MCP servers no longer re-announce full tool-name lists — they are summarized by server prefix
SDK hosts now receive a persistent localSettings suggestion for Bash permission prompts, so "Always allow" writes to .claude/settings.local.json
EnterWorktree now creates the new branch from local HEAD as documented, instead of origin/<default-branch> — unpushed commits are no longer dropped
Fixed crash loop when piping very large input (>10 MB) to claude -p via stdin, MCP tool results dropping images, and sessions on 1M-context models being falsely blocked with "Prompt is too long" before reaching the API limit

v2.1.126 (May 2026)

Added claude project purge [path] subcommand to delete all Claude Code state for a project (transcripts, tasks, file history, config entry) with --dry-run, -y/--yes, -i/--interactive, and --all
/model picker now lists models from your gateway's /v1/models endpoint when ANTHROPIC_BASE_URL points at an Anthropic-compatible gateway
--dangerously-skip-permissions now bypasses prompts for writes to .claude/, .git/, .vscode/, shell config files, and other previously-protected paths (catastrophic removal commands still prompt as a safety net)
claude auth login now accepts an OAuth code pasted into the terminal when the browser callback can't reach localhost (WSL2, SSH, containers)
OpenTelemetry: claude_code.skill_activated now fires for user-typed slash commands and carries a new invocation_trigger attribute (user-slash, claude-proactive, or nested-skill)
Auto mode: the spinner turns red when a permission check stalls, instead of looking like the tool is running
Windows: PowerShell 7 installed via the Microsoft Store, MSI without PATH, or .NET global tool is now detected. When the PowerShell tool is enabled, PowerShell becomes the primary shell instead of Bash
**Security:** Fixed allowManagedDomainsOnly / allowManagedReadPathsOnly being ignored when a higher-priority managed-settings source lacked a sandbox block
Fixed pasting an image larger than 2000px breaking the session — images are downscaled on paste, and oversized images in history are removed and the request retried
Fixed OAuth login failing with timeout on slow or proxied connections, in IPv6-only devcontainers, and when the browser callback can't reach localhost

v2.1.123 (April 2026)

Fixed OAuth authentication failing with a 401 retry loop when CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1 is set

v2.1.122 (April 2026)

Added ANTHROPIC_BEDROCK_SERVICE_TIER environment variable for Bedrock service tier selection (default / flex / priority, sent as X-Amzn-Bedrock-Service-Tier header)
Pasting a PR URL into the /resume search box now finds the session that created that PR (GitHub, GitHub Enterprise, GitLab, Bitbucket)
/mcp now shows claude.ai connectors hidden by a manually-added server with the same URL, with a hint to remove the duplicate
Clarified the /mcp message shown when an MCP server is still unauthorized after browser sign-in
OpenTelemetry: numeric attributes on api_request/api_error log events are now emitted as numbers, not strings
OpenTelemetry: added claude_code.at_mention log event for @-mention resolution
Fixed /branch producing forks that fail with "tool_use ids were found without tool_result blocks" from rewound timelines
Fixed /model not showing the Effort option for Bedrock application inference profile ARNs
Fixed Vertex AI / Bedrock returning output_config: Extra inputs are not permitted on session-title generation and other structured-output queries
Fixed ToolSearch missing MCP tools that connected after session start in nonblocking mode
Fixed !exit / !quit in bash mode terminating the CLI instead of running as a shell command
Fixed images sent to newer models being resized to 2576px per side instead of the correct 2000px maximum
Voice mode: keybindings bound to Caps Lock now show an error since terminals don't deliver Caps Lock as a key event

v2.1.121 (April 2026)

Added claude plugin prune to remove orphaned auto-installed plugin dependencies; plugin uninstall --prune cascades
Added alwaysLoad option to MCP server config — when true, all tools from that server skip tool-search deferral and are always available
Added a type-to-filter search box to /skills so you can find a skill in long lists without scrolling
PostToolUse hooks can now replace tool output for all tools via hookSpecificOutput.updatedToolOutput (previously MCP-only)
--dangerously-skip-permissions no longer prompts for writes to .claude/skills/, .claude/agents/, and .claude/commands/
/terminal-setup now enables iTerm2's "Applications in terminal may access clipboard" setting so /copy works, including from tmux
SDK and claude -p: CLAUDE_CODE_FORK_SUBAGENT=1 now works in non-interactive sessions
MCP servers that hit a transient error during startup now auto-retry up to 3 times instead of staying disconnected
Fullscreen mode: typing into the prompt no longer jumps scroll back to the bottom after you've scrolled up
Dialogs that overflow the terminal are now scrollable with arrow keys, PgUp/PgDn, home/end, and mouse wheel
The terminal tab session title is now generated in your configured language setting
Vertex AI: support X.509 certificate-based Workload Identity Federation (mTLS ADC)
OpenTelemetry: added stop_reason, gen_ai.response.finish_reasons, and user_system_prompt (gated behind OTEL_LOG_USER_PROMPTS) to LLM request spans
[VSCode] /context now opens a native token usage dialog

v2.1.120 (April 2026)

Added claude ultrareview [target] subcommand to run /ultrareview non-interactively from CI or scripts — prints findings to stdout (--json for raw output) and exits 0 on completion or 1 on failure
Windows: Git for Windows (Git Bash) is no longer required — when absent, Claude Code uses PowerShell as the shell tool
Skills can now reference the current effort level with ${CLAUDE_EFFORT} in their content
Set AI_AGENT environment variable for subprocesses so gh can attribute traffic to Claude Code
claude plugin validate now accepts $schema, version, and description at the top level of marketplace.json and $schema in plugin.json
Auto-compact in auto mode now displays auto (lowercase, no token count) instead of a misleading token value
Show a "use PgUp/PgDn to scroll" hint when the terminal sends arrow keys instead of scroll events
Faster session start when you have many claude.ai connectors configured but not authorized
[VSCode] /usage now opens the native Account & Usage dialog instead of returning plain-text session cost
[VSCode] Voice dictation now respects the language setting in ~/.claude/settings.json

v2.1.119 (April 2026)

/config settings (theme, editor mode, verbose, etc.) now persist to ~/.claude/settings.json and participate in project/local/policy override precedence
Hooks: PostToolUse and PostToolUseFailure hook inputs now include duration_ms (tool execution time, excluding permission prompts and PreToolUse hooks)
Subagent and SDK MCP server reconfiguration now connects servers in parallel instead of serially
Plugins pinned by another plugin's version constraint now auto-update to the highest satisfying git tag
--from-pr now accepts GitLab merge-request, Bitbucket pull-request, and GitHub Enterprise PR URLs
--print mode now honors the agent's tools: and disallowedTools: frontmatter, matching interactive-mode behavior
--agent <name> now honors the agent definition's permissionMode for built-in agents
PowerShell tool commands can now be auto-approved in permission mode, matching Bash behavior
Added prUrlTemplate setting to point the footer PR badge at a custom code-review URL
Added CLAUDE_CODE_HIDE_CWD environment variable to hide the working directory in the startup logo
OpenTelemetry: tool_result and tool_decision events now include tool_use_id; tool_result also includes tool_input_size_bytes
Status line: stdin JSON now includes effort.level and thinking.enabled
Security: blockedMarketplaces now correctly enforces hostPattern and pathPattern entries

v2.1.118 (April 2026)

Added vim visual mode (v) and visual-line mode (V) with selection, operators, and visual feedback
Merged /cost and /stats into /usage — both remain as typing shortcuts that open the relevant tab
Create and switch between named custom themes from /theme, hand-edit JSON files in ~/.claude/themes/; plugins can also ship themes via a themes/ directory
Hooks can now invoke MCP tools directly via type: "mcp_tool"
Added DISABLE_UPDATES env var to completely block all update paths including manual claude update — stricter than DISABLE_AUTOUPDATER
Auto mode: include "$defaults" in autoMode.allow, autoMode.soft_deny, or autoMode.environment to add custom rules alongside the built-in list instead of replacing it
Added claude plugin tag to create release git tags for plugins with version validation
--continue/--resume now find sessions that added the current directory via /add-dir
/color now syncs the session accent color to claude.ai/code when Remote Control is connected
Fixed /fork writing the full parent conversation to disk per fork — now writes a pointer and hydrates on read

v2.1.117 (April 2026)

/model selections persist across restarts even when the project pins a different model; the startup header shows when the active model comes from a project or managed-settings pin
/resume now offers to summarize stale, large sessions before re-reading them (matching the existing --resume behavior)
Native builds on macOS and Linux replace Glob and Grep with embedded bfs and ugrep available through the Bash tool — faster searches without a separate tool round-trip (Windows and npm-installed builds unchanged)
Default effort for Pro/Max subscribers on Opus 4.6 and Sonnet 4.6 is now high (was medium)
Forked subagents can now be enabled on external builds by setting CLAUDE_CODE_FORK_SUBAGENT=1
Agent frontmatter mcpServers are now loaded for main-thread agent sessions via --agent
Faster startup when both local and claude.ai MCP servers are configured (concurrent connect now default)
Plain-CLI OAuth sessions no longer die with "Please run /login" when tokens expire mid-session — tokens are now refreshed reactively on 401
Fixed Opus 4.7 sessions showing inflated /context percentages and autocompacting too early — now computes against Opus 4.7's native 1M context window

v2.1.113-v2.1.116 (April 2026)

CLI now spawns a platform-specific native Claude Code binary instead of bundled JavaScript
Added sandbox.network.deniedDomains setting to override allowedDomains wildcards and block specific domains
Hardened Bash(rm:*) / Bash(find:*) allow rules; deny rules now match commands wrapped in env / sudo / watch and similar exec wrappers
Fullscreen mode: Shift+↑/↓ now scrolls the viewport when extending a selection past the visible edge
Multiline input: Ctrl+A / Ctrl+E now move to the start/end of the current logical line (readline); Windows Ctrl+Backspace deletes the previous word
Improved /loop: Esc cancels pending wakeups; /ultrareview now launches faster with parallelized checks and a diffstat in the launch dialog
/extra-usage and @-file autocomplete are now available from Remote Control (mobile/web) clients
Fixed a crash in the permission dialog when an agent teams teammate requested tool permission (v2.1.114)
/resume is up to 67% faster on 40MB+ sessions; /doctor can now be opened while Claude is responding (v2.1.116)
Faster MCP startup when multiple stdio servers are configured; resources/templates/list is deferred until first @-mention (v2.1.116)
Sandbox auto-allow no longer bypasses the dangerous-path safety check for rm / rmdir targeting /, $HOME, or other critical system directories (v2.1.116)
Fixed Ctrl+- undo and Cmd+Left/Right line navigation in terminals using the Kitty keyboard protocol (iTerm2, Ghostty, kitty, WezTerm, Windows Terminal) (v2.1.116)

v2.1.108-v2.1.112 (April 2026)

Claude Opus 4.7 available with new xhigh effort level (between high and max)
Auto mode expanded to Max subscribers on Opus 4.7; --enable-auto-mode flag removed (use --permission-mode auto)
Session recap: one-line summary after returning from idle, plus on-demand /recap
Fullscreen rendering via /tui fullscreen — flicker-free alt-screen renderer with scrollback dump and editor integration
/focus toggles a minimal view showing only the last prompt and final response
Push notification tool — Claude can send mobile push alerts when Remote Control is enabled
New /ultrareview skill runs deep multi-agent cloud code review
New /less-permission-prompts skill proposes a prioritized Bash/MCP allowlist

v2.1.x (March 2026)

Remote Control: drive local sessions from browser or mobile
Channels (Telegram/Discord/iMessage) push events into running sessions
Agent Teams: coordinate multiple Claude Code instances working together
Voice dictation (Push-to-Talk) with 20-language support
Fast Mode: 2.5x speed boost for Opus 4.6
Plugin system to package and share skills, agents, hooks, and MCP servers
Skills (SKILL.md) for custom workflows, plus bundled /batch and /simplify
Git worktree isolation for parallel sessions without file conflicts
Scheduled tasks (cloud, desktop, GitHub Actions, /loop)
/btw side questions and prompt suggestions
Checkpointing & rewind to restore code/conversation to any point
Keybinding customization (~/.claude/keybindings.json)

v2.0.x (February 2026)

Claude Opus 4.6 model support
Claude Code on the Web (claude.ai/code) for browser-based execution
Improved MCP server stability and new tools
/diff command for git diff visualization
/memory command for memory management
Auto Memory for automatic cross-session learning

v1.0.x (June 2025)

File read/write, code generation, refactoring
Git operations (commit, PR creation, review)
Terminal command execution
CLAUDE.md for project configuration
Custom slash commands
MCP (Model Context Protocol) integration

About Claude Code Features Guide

The Claude Code Features Guide is a comprehensive reference that systematically organizes all features of Claude Code, Anthropic's AI-powered coding assistant for the terminal. Unlike command references that list syntax, this guide focuses on what you can actually accomplish — from code editing and Git operations to MCP server integration, sub-agent delegation, and team coordination. Features are organized into clear categories including Code Editing, Git & Version Control, Testing & Debugging, MCP Integration, Customization (CLAUDE.md, skills, hooks), Agents & Teams, and Advanced capabilities like headless mode and SDK integration. Search by keyword or filter by category to quickly find the feature you need.

Key Features

Comprehensive list of all features organized into 8 categories
Usage examples and related commands for each feature
Category filters and keyword search
Version-based update history to quickly find new features

Use Cases

Exploring what Claude Code can do before adopting it on a production codebase at work
Finding the right feature (code generation, test writing, refactoring, bug fixing) for a specific task without trial-and-error
Onboarding a new engineer to Claude Code by sharing a concise overview of agent mode, MCP integrations, and CLAUDE.md conventions
Checking which features support headless/CI use cases before wiring Claude Code into a GitHub Actions or GitLab CI pipeline
Comparing Claude Code's built-in capabilities against Cursor or Copilot when evaluating AI coding tools for your team

FAQ

What's the difference between the Features Guide and the Commands Reference?

The Commands Reference lists every CLI flag, slash command, and keyboard shortcut. The Features Guide is organized by capability — bug fixing, test generation, refactoring, MCP integrations — so you can find what Claude Code can do for a specific workflow without scanning command syntax.

How much does Claude Code cost?

Claude Code requires an Anthropic API key and charges based on token usage (input + output). There is also a Claude Pro/Team subscription that includes Claude Code access. This reference guide is free to use regardless.

What is MCP and why does it matter?

MCP (Model Context Protocol) lets Claude Code connect to external tools — GitHub, Jira, databases, Slack, browsers, and more. Instead of switching tabs, you can query a database or file a GitHub issue directly from your terminal session.

What is CLAUDE.md and should I commit it?

CLAUDE.md is a project-level config file that tells Claude Code your coding conventions, build commands, and team standards. Yes, commit it — every developer on the team will then get consistent Claude Code behavior without manual setup.

Can Claude Code read and edit files autonomously?

Yes. In agent mode, Claude Code can read multiple files, make edits, run tests, and iterate — all in one go. You can control how much autonomy it has via permission settings and the --dangerously-skip-permissions flag for fully automated pipelines.

How does Claude Code compare to GitHub Copilot or Cursor?

Claude Code is a terminal-based agent that can run shell commands and operate across your entire codebase, not just the current file. Copilot and Cursor focus on in-editor completions. Claude Code is better suited for multi-step tasks like refactoring, debugging, or writing tests across many files.

Claude Code Features Guide

How to Use

Search for a Feature

View Feature Details

Try It Out

Recent Updates

File Read & Write

Usage Examples

Multi-File Editing

Usage Examples

Code Generation

Usage Examples

Refactoring

Usage Examples

Bug Fixing

Usage Examples

Code Review

Usage Examples

Related Commands

Auto Lint Fix

Usage Examples

Commit Creation

Usage Examples

Related Commands

Pull Request Creation

Usage Examples

Related Commands

PR Review

Usage Examples

Related Commands

Branch Management

Usage Examples

Diff Analysis

Usage Examples

Related Commands

Conflict Resolution

Usage Examples

History Investigation

Usage Examples

Related Commands

Command Execution

Usage Examples

Related Commands

Environment Setup

Usage Examples

Related Commands

Package Management

Usage Examples

Build & Run

Usage Examples

Process Management

Usage Examples

Prompt Suggestions

Usage Examples

/btw (Side Questions)

Usage Examples

Related Commands

Session Recap

Usage Examples

Related Commands

Codebase Search

Usage Examples

Related Commands

Dependency Analysis

Usage Examples

Code Explanation

Usage Examples

Architecture Analysis

Usage Examples

Impact Analysis

Usage Examples

Image Input (Multimodal)

Usage Examples

Test Generation

Usage Examples

Test Execution & Debugging

Usage Examples

Test-Driven Development (TDD)

Usage Examples

Test Maintenance