Description

Comprehensive Claude Code CLI reference covering interactive mode, CLI flags, commands, environment variables, hooks, channels, checkpointing, plugins, and t...

README (SKILL.md)

Code with Claude

Name: code-with-claude
Author: awalesagar

Use this skill when the user asks about Claude Code CLI features, configuration, workflows, or best practices.

Claude Code Feature Map

Feature	What It Does	Reference
CLI flags	Launch-time options for sessions, models, permissions, prompts	`./references/cli-reference.md`
Commands	In-session `/` commands for model switching, permissions, context, workflows	`./references/commands.md`
Interactive mode	Keyboard shortcuts, Vim mode, multiline input, voice, task list	`./references/interactive-mode.md`
Hooks	Shell commands, HTTP endpoints, or LLM prompts at lifecycle events	`./references/hooks.md`
Plugins	Extend Claude Code with skills, agents, hooks, MCP/LSP servers	`./references/plugins.md`
Channels	MCP servers that push external events into a session	`./references/channels.md`
Checkpointing	Automatic file edit tracking with rewind and summarize	`./references/checkpointing.md`
Tools	Built-in tools: Bash, Edit, Read, Write, Agent, WebFetch, etc.	`./references/tools.md`
Environment variables	Control behavior via env vars or `settings.json`	`./references/env-vars.md`

Quick Reference: Essential CLI Flags

Flag	Purpose	Example
`--model`	Set model (alias or full name)	`claude --model opus`
`--continue`, `-c`	Resume most recent conversation	`claude -c`
`--resume`, `-r`	Resume by session ID or name	`claude -r "auth-refactor"`
`--print`, `-p`	Non-interactive print mode (SDK)	`claude -p "query"`
`--permission-mode`	Start in a specific permission mode	`claude --permission-mode plan`
`--worktree`, `-w`	Isolated git worktree session	`claude -w feature-auth`
`--bare`	Minimal mode, skip auto-discovery	`claude --bare -p "query"`
`--append-system-prompt`	Add to default system prompt	`claude --append-system-prompt "Use TS"`
`--effort`	Set effort level for session	`claude --effort high`

Quick Reference: Essential Commands

Command	Purpose
`/model [model]`	Switch model mid-session
`/compact [instructions]`	Compact conversation, free context
`/clear`	Clear conversation history
`/rewind`	Rewind to a checkpoint (Esc+Esc)
`/permissions`	Manage tool allow/deny rules
`/plan [description]`	Enter plan mode
`/batch \x3Cinstruction>`	Parallel large-scale changes
`/hooks`	View hook configurations
`/memory`	Edit CLAUDE.md memory files
`/diff`	Interactive diff viewer
`/cost`	Show token usage statistics
`/effort [level]`	Set effort (low/medium/high/max)

Quick Reference: Key Shortcuts

Shortcut	Action
`Ctrl+C`	Cancel current generation
`Esc+Esc`	Rewind / checkpoint menu
`Shift+Tab`	Cycle permission modes
`Ctrl+O`	Toggle transcript viewer
`Ctrl+B`	Background running tasks
`Ctrl+T`	Toggle task list
`Alt+P`	Switch model
`Alt+T`	Toggle extended thinking
`!command`	Run bash directly in session
`@path`	File path autocomplete

Quick Reference: Permission Modes

Mode	Behavior
`default`	Ask for each tool use
`acceptEdits`	Auto-approve file edits, ask for bash
`plan`	Read-only exploration, no writes
`auto`	AI classifier approves safe operations
`bypassPermissions`	Skip all permission prompts

Cycle modes with Shift+Tab. Start in a specific mode: claude --permission-mode plan.

Quick Reference: Core Tools

Tool	Description	Permission
`Bash`	Execute shell commands	Yes
`Read`	Read file contents	No
`Edit`	Targeted file edits	Yes
`Write`	Create or overwrite files	Yes
`Agent`	Spawn subagent with own context	No
`Glob`	Find files by pattern	No
`Grep`	Search file contents	No
`WebFetch`	Fetch URL content	Yes
`WebSearch`	Web search	Yes
`Monitor`	Background watch + react	Yes

Gotchas

Env vars don't persist across Bash tool commands. Use CLAUDE_ENV_FILE or a SessionStart hook.
Checkpoints don't track bash changes — only direct file edits via Edit/Write tools are tracked.
applyTo: "**" burns context — always-on instructions load on every interaction. Use specific globs.
macOS Alt shortcuts require Option-as-Meta in your terminal (iTerm2: Profiles → Keys → "Esc+").
--bare disables everything — hooks, skills, plugins, MCP, auto-memory, CLAUDE.md. Use for fast scripted calls.
Hook if patterns use tool permission syntax: Bash(rm *) matches bash commands starting with rm.
Plugin agents can't use hooks, mcpServers, or permissionMode frontmatter for security reasons.

When to Load References

Load the appropriate reference when the user needs detailed information:

CLI flags, system prompt customization, launch options → Read ./references/cli-reference.md
In-session commands, slash commands, skills list → Read ./references/commands.md
Keyboard shortcuts, Vim mode, multiline, voice, task list → Read ./references/interactive-mode.md
Hooks lifecycle, events, matchers, configuration → Read ./references/hooks.md
Plugins structure, manifest, MCP/LSP, distribution → Read ./references/plugins.md
Channels, webhook receivers, two-way chat bridges → Read ./references/channels.md
Checkpointing, rewind, restore, summarize → Read ./references/checkpointing.md
Built-in tools, Bash behavior, LSP, Monitor, PowerShell → Read ./references/tools.md
Environment variables, timeouts, model config, proxy → Read ./references/env-vars.md

Usage Guidance

This skill is documentation-only and appears coherent for its stated purpose. It does, however, document and include examples that can execute shell commands, start local servers, forward webhooks, or configure hooks that POST to HTTP endpoints. Before following any examples: (1) review any shell/HTTP commands and understand what they do; (2) do not paste secrets (API keys, OAuth tokens) into examples from untrusted sources; (3) be cautious enabling hooks or channels that open network listeners or relay permissions; and (4) treat provided example code as a template to review and harden rather than as safe-to-run copy/paste.

Capability Analysis

Type: OpenClaw Skill Name: code-with-claude Version: 1.0.0 The 'code-with-claude' skill bundle is a comprehensive documentation and reference guide for the Claude Code CLI tool. It consists of a SKILL.md file and several reference markdown files (e.g., cli-reference.md, tools.md, hooks.md) that provide detailed information on commands, environment variables, and configuration. The content is purely informational, designed to help an AI agent answer user questions about the tool's features. No malicious code, data exfiltration, or prompt injection attempts were identified.

Capability Tags

requires-oauth-token

Capability Assessment

✓ Purpose & Capability

The skill is a CLI/reference guide for Claude Code and the included files (CLI flags, hooks, plugins, channels, tools, env vars, etc.) match the name and description. It does not request unrelated credentials, binaries, or config paths.

ℹ Instruction Scope

The SKILL.md and referenced documents are purely documentation and do not instruct the agent to read arbitrary user files or exfiltrate data. However the references contain runnable examples (e.g., spawning an MCP server, hook command examples, and hook HTTP POSTs) and descriptions of powerful features (Bash tool, hooks that run shell commands, bypass permission modes). Those examples are expected in a CLI reference but could lead to risky actions if followed by a user or an agent without review.

✓ Install Mechanism

No install specification and no code files to execute — instruction-only skill. Nothing is downloaded or written to disk by the skill itself.

ℹ Credentials

The skill declares no required environment variables or credentials. The documentation lists many environment variables (e.g., ANTHROPIC_API_KEY, CLAUDE_CODE_OAUTH_TOKEN, CLAUDE_ENV_FILE) as part of product configuration; this is expected for a reference but users should not supply secrets to untrusted plugins or follow examples that persist tokens without review.

✓ Persistence & Privilege

always:false and user-invocable:true. The skill does not request permanent presence or elevated privileges and does not modify other skills or system-wide settings.

Version History

v1.0.0

- Initial release of the code-with-claude skill. - Provides a comprehensive Claude Code CLI feature reference, including commands, flags, environment variables, hooks, plugins, channels, checkpointing, and tools. - Includes quick reference sections for essential commands, keyboard shortcuts, permission modes, and core tools. - Highlights common gotchas, best practices, and when to load specific reference files. - Designed to assist users with configuration, workflows, and effective use of the Claude Code CLI.

Metadata

Slug code-with-claude

Version 1.0.0

License MIT-0

All-time Installs 0

Active Installs 0

Total Versions 1

Frequently Asked Questions

What is code-with-claude?

Comprehensive Claude Code CLI reference covering interactive mode, CLI flags, commands, environment variables, hooks, channels, checkpointing, plugins, and t... It is an AI Agent Skill for Claude Code / OpenClaw, with 124 downloads so far.

How do I install code-with-claude?

Run "/install code-with-claude" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is code-with-claude free?

Yes, code-with-claude is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does code-with-claude support?

code-with-claude is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created code-with-claude?

It is built and maintained by Sagar Awale (@awalesagar); the current version is v1.0.0.

More Skills

code-with-claude