Command Reference

Complete guide to Claude Octopus commands and workflows

Command Reference

Claude Octopus provides 46+ commands organized into workflow commands, system commands, and specialized skills. Commands can be invoked explicitly (e.g., /octo:embrace) or triggered naturally through conversational prompts.

Command Categories

Workflow Commands

Double Diamond methodology phases and full lifecycle workflows

System Commands

Setup, configuration, diagnostics, and mode switching

Skill Commands

Specialized capabilities like debate, review, security, TDD, and more

Quick Reference

Core Workflow Commands

Command	Purpose	Multi-AI	When to Use
/octo:embrace	Full 4-phase workflow	Yes	Complex features needing research → delivery
/octo:discover	Research & exploration	Yes	Starting new features, researching options
/octo:define	Requirements & scope	Yes	Clarifying requirements, scoping features
/octo:develop	Implementation	Yes	Building features with quality gates
/octo:deliver	Validation & review	Yes	Final quality assurance and testing

Alternative Phase Names

Each workflow phase has alternative names from the original Double Diamond methodology:

• Discover = Probe (divergent research)
• Define = Grasp (convergent scoping)
• Develop = Tangle (divergent implementation)
• Deliver = Ink (convergent validation)

Smart Router

Command	Purpose	Example
/octo	Natural language router	/octo research OAuth patterns

System Commands

Command	Purpose	Example
/octo:setup	Configure providers	Check setup status and get config instructions
/octo:doctor	Environment diagnostics	Verify all components working
/octo:status	Project state	View current workflow status
/octo:resume	Restore context	Resume from last session
/octo:dev	Dev Work mode	Switch to development personas
/octo:km	Knowledge mode	Switch to research personas

Specialized Skills

Command	Purpose	Multi-AI	When to Use
/octo:debate	AI debate hub	Yes	Comparing options, adversarial review
/octo:review	Code review	Yes	Pre-commit checks, quality assessment
/octo:security	Security audit	Yes	OWASP compliance, vulnerability detection
/octo:tdd	Test-driven dev	Yes	Red-green-refactor implementation
/octo:factory	Dark Factory	Yes	Spec-in, software-out autonomous pipeline
/octo:prd	PRD generation	Yes	AI-optimized product requirements
/octo:claw	OpenClaw admin	No	Gateway lifecycle management

Planning & Orchestration

Command	Purpose	Multi-AI	When to Use
/octo:plan	Strategic planning	No	Create execution plans before running
/octo:parallel	Team of teams	Yes	Decompose work into parallel packages
/octo:multi	Force multi-provider	Yes	Manual override for parallel execution
/octo:spec	NLSpec authoring	Yes	Write structured specifications

Command Syntax

Explicit Commands

/octo:embrace "build user authentication"
/octo:discover "OAuth security patterns"
/octo:review src/auth.ts

Natural Language (Smart Router)

/octo research OAuth patterns
/octo build authentication system
/octo should we use Redis or Memcached?

With Arguments

/octo:factory --spec path/to/spec.md
/octo:km on                    # Switch to Knowledge Work mode
/octo:km off                   # Switch to Dev Work mode

Common Flags and Options

Global Flags

--autonomy supervised    # Pause after each phase (default)
--autonomy semi-auto    # Only pause on quality gate failures
--autonomy autonomous   # Run all phases without intervention
--ci                   # Non-interactive mode for automation

Model Configuration

export OCTOPUS_CODEX_MODEL="gpt-4"           # Override Codex model
export OCTOPUS_GEMINI_MODEL="gemini-pro"     # Override Gemini model
export OCTOPUS_OPUS_MODE="fast"              # Use fast Opus (6x cost)

Provider Control

export CODEX_SKIP=true     # Disable Codex provider
export GEMINI_SKIP=true    # Disable Gemini provider

Visual Indicators

When multi-AI workflows execute, you’ll see provider indicators:

• 🐙 Claude Octopus mode active
• 🔴 Codex CLI executing (uses OPENAI_API_KEY)
• 🟡 Gemini CLI executing (uses GEMINI_API_KEY)
• 🟣 Perplexity web search (uses PERPLEXITY_API_KEY)
• 🔵 Claude processing (included with Claude Code)

Cost Awareness

Provider	Cost Range	Source
Codex	$0.01-0.15/query	User’s OPENAI_API_KEY
Gemini	$0.01-0.03/query	User’s GEMINI_API_KEY
Perplexity	$0.01-0.05/query	User’s PERPLEXITY_API_KEY
Claude	Included	Claude Code subscription
Opus Fast	$30/$150 per MTok	6x cost, lower latency

For simple tasks that don’t need multi-AI perspectives, Claude handles them directly without invoking external providers.

Auto-Activation

Claude Octopus automatically activates for:

• Research keywords: “research”, “investigate”, “explore”
• Build keywords: “build”, “create”, “implement” (with clear target)
• Validation keywords: “validate”, “review”, “check”, “audit”
• Debate keywords: “should”, “vs”, “or”, “compare”
• Lifecycle keywords: “complete”, “full”, “end-to-end”

No need to memorize commands—just describe what you want naturally.

Next Steps

Workflow Commands

Learn the Double Diamond phases

System Commands

Configure your environment

Skill Commands

Explore specialized capabilities