94 cards across 6 sections
CLAUDE.md
-p
/diff
/cost
--permission-prompts none
.claude/rules/
curl -fsSL https://claude.ai/install.sh | bash
install.ps1
install.cmd
brew install --cask claude-code
claude-code@latest
winget install Anthropic.ClaudeCode
claude
claude "do this task"
ANTHROPIC_API_KEY
claude -p "query"
claude --continue
-c
claude --resume
-r
/resume
claude --worktree <name>
/doctor
~/.claude/CLAUDE.md
./CLAUDE.md
./.claude/CLAUDE.md
./CLAUDE.local.md
CLAUDE.local.md
@path/to/file
.claude/rules/*.md
paths:
~/.claude/projects/<project>/memory/
user
feedback
project
reference
MEMORY.md
/init
/memory
/clear
/compact
/config
/context
/mcp
/agents
/hooks
/permissions
/model
.claude/commands/
/command-name
description
arguments
SKILL.md
~/.claude/skills/<name>/SKILL.md
.claude/skills/<name>/SKILL.md
/skill-name [args]
disable-model-invocation: true
/
!
syntax inside a
runs a shell command before Claude sees the skill and injects the live output, useful for pulling in a current
or
.claude/agents/
~/.claude/agents/
name
tools
model
haiku
sonnet
opus
inherit
permissionMode
maxTurns
skills
memory
@"name (agent)"
claude --agent name
claude mcp add --transport <http|stdio|sse> <name> <url-or-command>
--
--scope local
~/.claude.json
--scope project
.mcp.json
--scope user
claude mcp list
claude mcp get <name>
claude mcp remove <name>
PreToolUse
PostToolUse
UserPromptSubmit
Stop
SubagentStop
SessionStart
SessionEnd
hooks
settings.json
matcher
Bash
Edit|Write
mcp__.*
2
0
default
acceptEdits
plan
auto
dontAsk
bypassPermissions
Tool
Tool(specifier)
Bash(npm run *)
WebFetch(domain:example.com)
allow
ask
deny
.claude/settings.local.json
Shift+Tab
plan mode on
/plan
claude --permission-mode plan
/compact [instructions]
/autocompact <tokens>
--settings
.claude/settings.json
~/.claude/settings.json
/status
gh pr create
/code-review
/review
--fix
--comment
claude-code-action
/install-github-app
@claude
prompt
@
Cmd+Esc
Ctrl+Esc
/terminal-setup
claude -p "query" --output-format json
--allowedTools
--permission-mode
--max-turns
cat file | claude -p "..."
claude> /init
README
package.json
See @README for an overviewand @package.json for scripts.
npm run *
git commit *
{ "permissions": { "allow": ["Bash(npm run *)"] }}
use a subagent to investigate howauth handles token refresh
claude mcp add --scope project \ --transport http docs https://x/mcp
rm -rf
"PreToolUse": [{ "matcher": "Bash", "hooks": [{ "type": "command", "command": "./hooks/block-rm.sh" }]}]
/compact focus on the auth bug fix
@"code-reviewer (agent)" review this
tail -200 app.log | \ claude -p "flag any anomalies"
--fix --comment
/code-review --fix --comment
--worktree
claude --worktree feature-auth
/skill-name
---disable-model-invocation: true---
--continue
--resume
claude --continueclaude --resume
defaultMode: "plan"
{ "defaultMode": "plan" }
claude mcp add
--print
claude -p "task" --output-format json --allowedTools "..." --permission-mode dontAsk
--max-budget-usd
.git
.claude
Bash(git * main)
Bash(git *)
git push
Bash(git commit *)
claude mcp add --transport stdio <name> -- <command> <args>
headers
env
${VAR}
.env
Bash(command:rm *)
Bash(rm *)