1 min read
Terminal CLI - Claude Code
The primary command-line entry point for Claude Code sessions.
GPT-6 In 7 Minutes
Reference Hub
Commands, tools, hooks, MCP, subagents, teams, permissions, and workflow details organized into one practical reference.

17 entries
Core workflow, modes, shortcuts, and day-one setup.
1 min read
The primary command-line entry point for Claude Code sessions.
1 min read
Real-time prompt loop with history, completions, and multiline input.
1 min read
50+ customizable shortcuts for cancel, history, transcript, and more.
1 min read
Approve each action manually - the safest mode for new tasks.
1 min read
Explore and propose changes without executing them.
1 min read
Skip all permission checks. Container and VM use only.
1 min read
Granular allow/ask/deny rules per tool with wildcard patterns.
1 min read
Filesystem and network isolation for Bash tool calls on Linux and macOS.
1 min read
Use opus, sonnet, haiku, and best to switch models easily.
1 min read
Hybrid mode: Opus for planning, Sonnet for execution.
1 min read
Low, medium, high, xhigh, and max for adaptive reasoning control.
1 min read
Interactive UI to switch models and effort sliders mid-session.
1 min read
Add gateway or custom models to the picker via environment variables.
1 min read
Hide skills from Claude's auto-selection until manually triggered.
1 min read
Fires when a permission dialog appears. Auto-approve or auto-deny.
1 min read
Fires when auto mode or a rule denies an action.
1 min read
Require lead approval before teammates execute their tasks.
21 entries
Bash, edit, read, grep, glob, notebook, LSP, and command behavior.
1 min read
Full vim keybindings (normal and insert modes) for prompt editing.
1 min read
Prefix prompts with ! to run shell commands directly, bypassing Claude.
1 min read
Read file contents with line limiting, offset, and binary support.
1 min read
Create or overwrite files; requires permission for existing paths.
1 min read
Targeted edits to specific sections without rewriting entire files.
1 min read
Batch edit multiple files in a single atomic operation.
1 min read
Modify Jupyter notebook cells directly without touching JSON.
1 min read
File discovery via pattern matching across the repository.
1 min read
Search file contents by pattern with regex support.
1 min read
Jump to definitions, find references, and type-check via language servers.
1 min read
Execute shell commands with persistent working directory in project bounds.
1 min read
Native PowerShell execution on Windows and optionally Unix hosts.
1 min read
Background monitoring of logs, files, and long-running processes.
1 min read
Fetch and parse content from URLs, including JS-rendered pages.
1 min read
Perform web searches and return ranked results with snippets.
1 min read
Auto-approve file edits and common filesystem commands.
1 min read
Limit which tools a subagent can access.
1 min read
Fires before any tool executes. Allow, deny, defer, or modify the call.
1 min read
Fires after a successful tool call. Good for feedback and follow-ups.
1 min read
Fires on tool execution errors for logging, alerting, and retry.
1 min read
Deferred tool loading reduces context overhead for large MCP suites.
20 entries
Lifecycle events, approvals, prompts, compacting, and automation points.
1 min read
Context-aware follow-up suggestions derived from git history.
1 min read
Automatic reuse of cached context for substantial cost reduction.
1 min read
Background context summarization when the window starts filling up.
1 min read
Event-driven automation with 20+ lifecycle events.
1 min read
Fires when a session begins; load env vars and initialize state.
1 min read
Fires when a session terminates.
1 min read
Fires before Claude processes user input; can validate or block.
1 min read
Fires when a slash command expands; can block or inject context.
1 min read
Fire when subagents spawn and finish.
1 min read
Fire on task lifecycle events.
1 min read
Fires when Claude finishes responding. Can prevent the stop.
1 min read
Fires when watched files change on disk.
1 min read
Fire when settings or CLAUDE.md files change during a session.
1 min read
Fire before and after context compaction.
1 min read
Fires when an MCP server requests input from the user.
1 min read
Run shell scripts on events with environment variable passing.
1 min read
Use Claude itself to handle hook logic instead of shell scripts.
1 min read
Spawn subagents to handle complex hook logic.
1 min read
Run hooks in the background without blocking the session.
1 min read
Execute MCP prompts as commands via the slash menu.
7 entries
Servers, resources, prompts, OAuth, scopes, and managed MCP setup.
1 min read
Route specific MCP servers only to specific subagents.
1 min read
Connect external tools and data sources via the open MCP standard.
1 min read
Local, project, user, and plugin-level MCP configurations.
1 min read
Pre-configured or dynamic OAuth for remote MCP servers.
1 min read
Reference and read resources exposed by MCP servers.
1 min read
Receive push messages from MCP servers via channels.
1 min read
Admin-controlled allow and deny lists for MCP servers.
11 entries
Definitions, tools, memory, context isolation, and team patterns.
1 min read
Run a skill in an isolated context via fork mode.
1 min read
Spawn isolated workers with independent context windows.
1 min read
Researcher, auditor, reviewer, and other ready-made subagent types.
1 min read
Create reusable subagent definitions at project or user level.
1 min read
Configure model, tools, MCP, skills, memory, and scoping.
1 min read
Auto-memory that persists across multiple subagent invocations.
1 min read
Prevent bloating the main conversation with research or exploration.
1 min read
Continue a subagent's work across sessions.
1 min read
Coordinate multiple Claude Code instances with a shared task list.
1 min read
Coordinator agent that assigns tasks and synthesizes findings.
1 min read
Reuse custom subagent types as Agent Teams members.
11 entries
Skills, frontmatter, invocation, auto-memory, and reusable context.
1 min read
Persistent project instructions loaded every session; supports nested dirs.
1 min read
Path-specific rules that only load for matching files.
1 min read
Automatic session-to-session memory of build commands, errors, and learnings.
1 min read
View and edit auto-memory and CLAUDE.md via the /memory command.
1 min read
Reusable markdown files with instructions and workflows.
1 min read
/simplify, /batch, /debug, /fast, and other built-in skills.
1 min read
Trigger with /skillname or let Claude auto-load when relevant.
1 min read
Configure model, effort, tools, MCP servers, and invocation scope.
1 min read
Pass arguments to skills with string substitution support.
1 min read
Pre-approve tools before a skill executes so it runs without prompts.
1 min read
Changes to skill files are detected and reloaded automatically.
25 entries
Everything else that does not fit a tighter section.
1 min read
Run Bash commands with Ctrl+B and retrieve output by task ID.
1 min read
Per-directory prompt history with Ctrl+R reverse search.
1 min read
Hold-to-record voice input on macOS, Linux, and Windows.
1 min read
Shift+Enter, Option+Enter, or backslash+Enter for multi-line prompts.
1 min read
Ask quick side questions without derailing the main task.
1 min read
SessionStart hooks can persist env vars across Bash tool calls.
1 min read
Stage, commit, branch, and open PRs without leaving the session.
1 min read
Isolated git worktrees for parallel Claude Code sessions.
1 min read
Clickable PR link in the footer with review state color coding.
1 min read
Full GitHub CLI support for automated PR and issue workflows.
1 min read
Define custom subagent types within your project's memory layer.
1 min read
Interactive timeline showing what's in context at each turn.
1 min read
Eliminate prompts with a background classifier that judges safety.
1 min read
Only pre-approved tools allowed. Fully non-interactive for CI.
1 min read
Auto-guarded directories like .git, .claude, and .vscode.
1 min read
Extended context window for Opus and Sonnet on supported plans.
1 min read
2.5x faster Opus at a higher token cost (research preview).
1 min read
Toggle with Alt+T. Claude reasons through complex problems before responding.
1 min read
Teammates claim and complete work independently from one list.
1 min read
Teammates communicate directly without routing through the lead.
1 min read
Run each teammate in its own tmux or iTerm2 pane.
1 min read
Pending tasks depend on others and unblock automatically.
1 min read
Run a prompt repeatedly on a fixed interval or self-paced.
1 min read
GUI-based scheduling on your local machine for recurring work.
1 min read
Managed scheduling on Anthropic infrastructure with API and GitHub triggers.

New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.