Claude Code Multi-Provider - Quick Command Reference

Reading time: 10 minutes | Skill level: Beginner | Version: v1.4.0 | Last updated: 2026-01-23

📋 All Available Commands

Provider Switching

Command	Provider	Description
`ccd`	Anthropic Direct	Best quality, production-ready
`ccc`	GitHub Copilot	Free with subscription
`cco`	Ollama Local	100% private, offline
`ccs`	Status	Check all providers

Model Selection (Copilot)

Command	Model	Speed	Use Case
`ccc-opus`	Claude Opus 4.6	Slow	Best quality, code review
`ccc-sonnet`	Claude Sonnet 4.6	Medium	Balanced (default)
`ccc-haiku`	Claude Haiku 4.5	Fast	Quick questions
`ccc-gpt`	GPT-4.1	Fast	Alternative perspective (0x quota)

Dynamic Model Selection

Copilot (25+ models available):

COPILOT_MODEL=claude-opus-4-6 ccc
COPILOT_MODEL=claude-haiku-4.5 ccc
COPILOT_MODEL=gpt-4.1 ccc
COPILOT_MODEL=gemini-3-pro-preview ccc

Ollama (local models):

OLLAMA_MODEL=devstral-small-2 cco       # Best agentic (default)
OLLAMA_MODEL=ibm/granite4:small-h cco   # Long context, less VRAM
OLLAMA_MODEL=qwen3-coder:30b cco        # Highest accuracy

Help & Version

Command	Output
`claude-switch --help`	Full usage guide
`claude-switch --version`	Version: v1.1.0
`ollama-check.sh --help`	Diagnostic help
`ollama-check.sh --version`	Version: v1.1.0
`ollama-optimize.sh --help`	Optimization help
`ollama-optimize.sh --version`	Version: v1.1.0

Diagnostic & Optimization

Command	Purpose
`ollama-check.sh`	11-point installation diagnostic
`ollama-optimize.sh`	Apply Apple Silicon optimizations
`ccs`	Check all provider status

Session Logs

Command	Description
`cat ~/.claude/claude-switch.log`	View all sessions
`tail -20 ~/.claude/claude-switch.log`	Recent sessions
`grep "mode=copilot" ~/.claude/claude-switch.log`	Filter by provider
`grep "Session ended" ~/.claude/claude-switch.log`	View durations

🚀 Quick Start Examples

1. Check Provider Status

ccs

2. Use Anthropic Direct

ccd
> Implement user authentication

3. Use Copilot (Free)

ccc
> Refactor this function for readability

4. Use Ollama (Private)

cco
> Analyze this proprietary algorithm

5. Quick Question with Haiku

ccc-haiku
> What's the difference between map and flatMap?

6. Code Review with Opus

ccc-opus
> Review this code for security vulnerabilities

7. Alternative with GPT

ccc-gpt
> Generate test cases for this function

🔧 Diagnostic Examples

Full Installation Check

ollama-check.sh

Expected output:

═══════════════════════════════════════════════════════════
  RÉSUMÉ
═══════════════════════════════════════════════════════════

   Ollama actif:      ✅
   API accessible:    ✅
   Modèle 32B:        ✅
   claude-switch:     ✅

Apply Optimizations

ollama-optimize.sh

Result: 26-39 tok/s on M4 Pro 48GB with Qwen2.5-Coder-32B

📊 Usage Patterns

Cost Optimization Workflow

# Morning: Explore (free)
ccc-haiku
> Help me understand this React project

# Afternoon: Implement (free)
ccc
> Add JWT authentication

# Before commit: Review (paid, best quality)
ccd
> Final security review

Privacy-First Workflow

# Public code: Use Copilot
ccc
> Implement standard REST API

# Proprietary code: Use Ollama
cco
> Optimize this trade secret algorithm

Speed-Optimized Workflow

# Quick iteration: Haiku
ccc-haiku
> 10 different approaches

# Refinement: Sonnet
ccc-sonnet
> Implement approach #3

# Polish: Opus
ccc-opus
> Final quality check

🎯 Strategic Command Selection

Scenario	Recommended Command	Reasoning
Production deployment	`ccd` or `ccc-opus`	Best quality
PR review	`ccd` or `ccc-opus`	Critical decisions
Learning new framework	`ccc` or `ccc-haiku`	Fast iteration, free
Prototyping MVP	`ccc`	Cost-effective
Proprietary code	`cco`	100% private
Offline work	`cco`	No internet required
Quick questions	`ccc-haiku`	Fastest responses
Comparing approaches	`ccc` then `ccc-gpt`	Different perspectives

📝 Session Management

Start Session

ccc              # Starts new session
ccc -c           # Resume previous session

View Session History

# Last 5 sessions
tail -5 ~/.claude/claude-switch.log

# Sessions with durations
grep "Session ended" ~/.claude/claude-switch.log

# Today's sessions
grep "$(date '+%Y-%m-%d')" ~/.claude/claude-switch.log

Filter by Provider

grep "mode=direct" ~/.claude/claude-switch.log    # Anthropic
grep "mode=copilot" ~/.claude/claude-switch.log   # Copilot
grep "mode=ollama" ~/.claude/claude-switch.log    # Ollama

🔄 Switching Between Providers Mid-Work

Not supported within same session. Exit and restart:

# Terminal 1: Start with Copilot
ccc
> Start implementing feature
> /exit

# Switch to Anthropic for critical review
ccd
> Review the code I just wrote

Alternative: Use multiple terminals in parallel:

# Terminal 1
ccc              # Development

# Terminal 2
cco              # Local testing

# Terminal 3
ccd              # Final review

💡 Pro Tips

1. Model Experimentation

Test same prompt with different models:

echo "Write a binary search function" | ccc-haiku
echo "Write a binary search function" | ccc-sonnet
echo "Write a binary search function" | ccc-opus
echo "Write a binary search function" | ccc-gpt

2. Cost Tracking

Track Anthropic usage vs Copilot:

# Count Anthropic sessions
grep "mode=direct" ~/.claude/claude-switch.log | wc -l

# Count Copilot sessions (free)
grep "mode=copilot" ~/.claude/claude-switch.log | wc -l

3. Performance Monitoring

Track Ollama performance:

# During session, in another terminal
watch -n 2 'ps aux | grep ollama | grep -v grep | awk "{printf \"RAM: %.2f GB\n\", \$6/1024/1024}"'

4. Model Aliases in Scripts

# Add to project-specific .envrc or Makefile
export COPILOT_MODEL="claude-opus-4-6"
alias review="ccc"  # Now uses Opus for this project

🆘 Troubleshooting Commands

Provider Not Working

ccs              # Check status first

copilot-api Not Running

copilot-api start

Ollama Issues

ollama-check.sh  # Full diagnostic
brew services restart ollama

Clear Logs (if too large)

tail -1000 ~/.claude/claude-switch.log > ~/.claude/claude-switch.log.tmp
mv ~/.claude/claude-switch.log.tmp ~/.claude/claude-switch.log

📚 Documentation Commands

Quick Reference

claude-switch --help          # Full help

Detailed Guides

# Read documentation
cat examples/multi-provider/README.md
cat examples/multi-provider/QUICKSTART.md
cat examples/multi-provider/OPTIMISATION-M4-PRO.md

All commands tested and working on macOS (M4 Pro 48GB) ✅

📚 Related Documentation

Quick Start Guide - Get started in 5 minutes
Cheatsheet - One-page printable reference
Decision Trees - Choose the right command
Model Switching Guide - Dynamic model selection
Best Practices - Strategic usage patterns

Back to: Documentation Index | Main README

FilesExpand file tree

COMMANDS.md

Latest commit

History