🛠️ Terminal AI Toolkit

Supercharge your terminal with AI-powered tools and scripts

---

📑 Table of Contents

• 🌟 Overview
• 📜 Scripts
• 🆓 Free API Providers
  • 💎 Gemini API
  • 🐙 GitHub Models
    • 🤖 GitHub Copilot Models
    • 🛒 GitHub Market Models
  • 🔀 OpenRouter
  • ⚡ Groq
  • 🚀 NVIDIA Build
  • 🦙 Ollama Cloud Models
• 💻 Local Model Providers
  • 🦙 Ollama
  • 🧠 LM Studio
• 🔀 API Proxies
  • 🔄 Claude Code Router
  • 🌉 Copilot API Bridge
• 📚 Detailed Tool Guides
• 🖥️ AI-Enhanced Terminals
  • ⚡ Warp Terminal
  • 🌊 Wave Terminal
  • 📟 iTerm2 AI
  • 🧩 TmuxAI

---

🌟 Overview

This repository is your comprehensive guide to getting the most out of AI tools in your terminal. It contains curated scripts, expert tips, and detailed guides for terminal-based AI development.

💡 Pro Tip: This is a companion to the awesome-terminals-ai list—your one-stop resource for terminal AI tools!

---

📜 Scripts

Useful scripts to enhance your AI terminal workflow:

Script	Description	Guide
📊 copilot-usage.sh	Check your GitHub Copilot usage and quota	Copilot CLI Guide
🤖 run-claude-copilot.sh	Run Claude Code with GitHub Copilot models	See below ⬇️

---

🆓 Free API Providers

💎 Gemini API

Access powerful Google Gemini models with generous free tier limits:

Feature	Gemini 2.5 Pro (Free)	Gemini 2.5 Flash (Free)
⚡ Rate Limit	2 requests/minute	15 requests/minute
📅 Daily Limit	50 requests/day	1,500 requests/day

• 📚 Rate Limits Documentation
• 🔑 Create API Key

🐙 GitHub Models

GitHub provides two types of AI model access for developers:

• 🤖 GitHub Copilot Models
• 🛒 GitHub Market Models

🤖 GitHub Copilot Models

Overview:

• 🌐 Endpoint: https://api.githubcopilot.com
• 📖 Documentation: Supported Models
• ⚡ Rate Limits: see Individual Plan Comparison

Premium request limits (per month):

Feature	GitHub Copilot Free	GitHub Copilot Pro	GitHub Copilot Pro+
Premium requests	0 per month	300 per month	1,500 per month

ℹ️ Exact limits and availability may change over time—always confirm via the official docs above.

Model multipliers:

• 📖 Model Multipliers Documentation
• Models (accessible via API) with a 0× multiplier for non-free plans (not counted toward premium usage): gpt-4.1, gpt-5-mini, gpt-4o

⚙️ Note: Some models need to be enabled at GitHub Copilot Features Settings before they become available for use.

⚠️ Integration Note: The endpoint https://api.githubcopilot.com supports OpenAI-compatible interface with GitHub OAuth Access Token (prefixed in gho_). However, the open-source proxy 🌉 Copilot API Bridge, authenticated with GitHub User Access Token (prefixed in ghu_), provides both OpenAI and Anthropic compatible interfaces.

List available models:

curl -L \
  -H "Accept: application/vnd.github+json" \
  -H "Authorization: Bearer ${OAUTH_TOKEN}" \
  https://api.githubcopilot.com/models | jq -r '.data[].id'

🛒 GitHub Market Models

Overview:

• 🌐 Endpoint: https://models.github.ai/inference
• 🔍 Browse: GitHub Marketplace Models
• 📊 Rate Limits: 4k input tokens, 4k output tokens per request

List available models:

curl -L \
  -H "Accept: application/vnd.github+json" \
  -H "Authorization: Bearer ${OAUTH_TOKEN}" \
  -H "X-GitHub-Api-Version: 2022-11-28" \
  https://models.github.ai/catalog/models | jq -r '.[].id'

🔀 OpenRouter

OpenRouter provides unified API access to multiple AI models—try different models using one API to find your best fit!

🔍 Browse Free Models

🆓 Free Models Available

Model	Link
GPT OSS 20B	Try it
GPT OSS 120B	Try it
GLM 4.5 Air	Try it
Qwen3 Next 80B A3B Instruct	Try it
Nemotron 3 Super	Try it
MiniMax M2.5	Try it
Gemma 4 26B A4B	Try it

Setup: 🔑 Generate API Key

💡 Rate Limits:

• With 10+ credits purchased: 1,000 requests/day
• Otherwise: 50 requests/day

⚡ Groq

Groq offers high-speed inference with free tier access.

🆓 Free Models

Available models from Rate Limits documentation:

• openai/gpt-oss-120b
• openai/gpt-oss-20b
• qwen/qwen3-32b
• moonshotai/kimi-k2-instruct-0905

Setup: 🔑 Generate API Key

🚀 NVIDIA Build

NVIDIA Build provides free API access to a wide selection of AI models optimized on NVIDIA infrastructure.

🆓 Example Models Available

Model	Full Model Name	Link
Qwen3.5 397B-A17B	qwen/qwen3.5-397b-a17b	Try it
MiniMax M2.5	minimaxai/minimax-m2.5	Try it
Kimi K2.5	moonshotai/kimi-k2.5	Try it
GLM-5	z-ai/glm5	Try it
GPT-OSS 120B	openai/gpt-oss-120b	Try it
DeepSeek V3.2	deepseek-ai/deepseek-v3_2	Try it

Setup:

• 🔑 Generate API Key
• 🔍 Browse All Models

💡 Note: Use the full model name (with namespace) when making API requests.

🦙 Ollama Cloud Models

Ollama now provides cloud-hosted models via API access, offering powerful AI capabilities without the need for local infrastructure. These models are accessible through a simple API and integrate seamlessly with popular AI coding tools.

💰 Pricing:

• 🆓 Free Plan - Available with hourly and daily usage limits
• 📈 Pay-per-use - No upfront costs or hardware investment required

🆓 Available Cloud Models

🔍 Search all cloud models

Model	Full Name	Use Case
⚡ GLM 5.1	glm-5.1	Next-gen flagship model for agentic engineering with state-of-the-art coding capabilities
🎯 MiniMax M2.7	minimax-m2.7	MiniMax's M2-series model for coding, agentic workflows, and professional productivity
💎 Gemma 4	gemma4	Multimodal model by Google DeepMind with advanced reasoning and coding capabilities
💻 Qwen3.5	qwen3.5	Multimodal models delivering exceptional utility and performance
🚀 Nemotron 3 Super	nemotron-3-super	120B open MoE model activating 12B params for efficient multi-agent applications
🔥 GPT-OSS	gpt-oss	Powerful reasoning, agentic tasks, and versatile developer use cases

🔗 Integration with AI Coding Tools

Ollama Cloud Models integrate seamlessly with popular AI coding tools and IDEs through native integrations and OpenAI-compatible APIs:

🎯 Supported AI Coding Tools & IDEs:

Tool	Integration Type	Documentation
VS Code	Native Extension	View Guide
JetBrains	Native Plugin	View Guide
Codex	API Integration	View Guide
Cline	API Integration	View Guide
Droid	API Integration	View Guide
Goose	API Integration	View Guide
Zed	Native Extension	View Guide

Key Benefits:

• OpenAI-compatible API - Use existing OpenAI client libraries
• Direct terminal integration - Run queries from command line
• No local setup required - Access powerful models via API
• Cost-effective - Pay-per-use without hardware investment
• Zero local storage - Models run in the cloud

Example API Usage:

# Query via REST API
curl https://api.ollama.ai/v1/chat/completions \
  -H "Authorization: Bearer ${OLLAMA_API_KEY}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3-coder:480b",
    "messages": [
      {"role": "user", "content": "Write a Python function to parse JSON"}
    ]
  }'

Setup:

• 🔑 Generate API Key
• 📚 View Documentation
• 🆓 Free Plan Available - Includes hourly and daily usage limits

💡 Pro Tip: Most integrations support both local and cloud models. For cloud models, append -cloud to the model name in your tool's configuration.

---

💻 Local Model Providers

🦙 Ollama

Ollama - Lightweight framework for running LLMs locally via command line.

Key Features:

• ⚡ Simple CLI interface
• 🌐 RESTful API
• 🐳 Docker-like model management
• 🤖 Popular models: LLaMA, Gemma, DeepSeek
• 🔌 OpenAI-compatible API
• 🖥️ Cross-platform support

☁️ Access to Cloud Models

Ollama also provides access to cloud-hosted models via the ollama command. Simply append -cloud (or :cloud for some models) to the model name:

# Example: Run a cloud-hosted model
ollama run qwen3-coder:480b-cloud

For details, see the 🦙 Ollama Cloud Models section.

📊 Ollama Model Performance

Model Sizes:

Model	Size
gpt-oss:120b	65 GB
gpt-oss:20b	13 GB
qwen3:8b	5.2 GB
qwen3:30b	18 GB

Performance Benchmark (tokens/second):

Machine	gpt-oss:120b	gpt-oss:20b	qwen3:8b	qwen3:30b
🖥️ Windows PC (Intel i9)	-	15 t/s	12 t/s	22 t/s
💻 MacBook Pro (M3 Max)	-	70 t/s	57 t/s	74 t/s
🖥️ Linux Server (Dual RTX 4090)	36 t/s	156 t/s	140 t/s	163 t/s

📋 Machine Specifications

• Windows PC (Intel i9):

  • CPU: Intel i9-12900
  • GPU: Intel UHD Graphics 770 (2 GB)
  • RAM: 64 GB

• MacBook Pro (M3 Max):

  • Apple M3 Max with 64 GB RAM

• Linux Server (Dual RTX 4090):

  • CPU: Xeon(R) w7-3445 (40 CPUs)
  • GPU: 2 × Nvidia RTX 4090
  • RAM: 128 GB

🖥️ LM Studio

LM Studio - User-friendly desktop GUI for running local LLMs with no technical setup required.

Key Features:

• 🛍️ Model marketplace
• 🌐 OpenAI-compatible API server
• 💬 Chat interface
• 📦 GGUF model support
• 💰 Free for personal & commercial use

---

🔀 API Proxies

Most AI tools support OpenAI-compatible APIs. For tools requiring Anthropic-compatible APIs, these solutions provide compatibility:

🔄 Claude Code Router

Claude Code Router - Routes Claude Code requests to different models with request customization.

📦 Installation (Linux/macOS)

# Install Claude Code CLI (prerequisite)
npm install -g @anthropic-ai/claude-code

# Install Claude Code Router
npm install -g @musistudio/claude-code-router

⚙️ Configuration Examples

Create ~/.claude-code-router/config.json with your preferred providers:

{
  "LOG": true,
  "API_TIMEOUT_MS": 600000,
  "Providers": [
    {
      "name": "gemini",
      "api_base_url": "https://generativelanguage.googleapis.com/v1beta/models/",
      "api_key": "$GEMINI_API_KEY",
      "models": ["gemini-2.5-flash", "gemini-2.5-pro"],
      "transformer": { "use": ["gemini"] }
    },
    {
      "name": "openrouter",
      "api_base_url": "https://openrouter.ai/api/v1/chat/completions",
      "api_key": "$OPENROUTER_API_KEY",
      "models": ["google/gemini-2.5-pro-preview", "anthropic/claude-sonnet-4"],
      "transformer": { "use": ["openrouter"] }
    },
    {
      "name": "grok",
      "api_base_url": "https://api.x.ai/v1/chat/completions",
      "api_key": "$GROK_API_KEY",
      "models": ["grok-beta"]
    },
    {
      "name": "github-copilot",
      "api_base_url": "https://api.githubcopilot.com/chat/completions",
      "api_key": "$GITHUB_TOKEN",
      "models": ["gpt-4o", "claude-3-7-sonnet", "o1-preview"]
    },
    {
      "name": "github-marketplace",
      "api_base_url": "https://models.github.ai/inference/chat/completions",
      "api_key": "$GITHUB_TOKEN",
      "models": ["openai/gpt-4o", "openai/o1-preview", "xai/grok-3"]
    },
    {
      "name": "ollama",
      "api_base_url": "http://localhost:11434/v1/chat/completions",
      "api_key": "ollama",
      "models": ["qwen3:30b", "gpt-oss:20b", "llama3.2:latest"]
    }
  ],
  "Router": {
    "default": "gemini,gemini-2.5-flash",
    "background": "ollama,qwen3:30b",
    "longContext": "openrouter,google/gemini-2.5-pro-preview"
  }
}

💻 Usage Commands

# Start Claude Code with router
ccr code

# Use UI mode for configuration
ccr ui

# Restart after config changes
ccr restart

# Switch models dynamically in Claude Code
/model ollama,llama3.2:latest

⚠️ Known Issue: The proxy for Ollama models does not work properly with Claude Code.

🌉 Copilot API Bridge

The GitHub Copilot API (https://api.githubcopilot.com) supports OpenAI-compatible interface with GitHub OAuth Access Token (prefixed in gho_). copilot‑api, an open‑source proxy authenticated with GitHub User Access Token (prefixed in ghu_), provides the necessary bridge: it exposes an OpenAI‑compatible interface as well as an Anthropic‑compatible interface, at the endpoint https://localhost:4141.

Installation and Authentication:

# Install copilot-api globally
npm install -g copilot-api

# Device authentication
copilot-api auth

# Start the API proxy
copilot-api start

The copilot-api tool is also available in specialized environments like the modern-linuxtools Singularity image on CVMFS.

CVMFS Setup:

# Setup the environment
source /cvmfs/atlas.sdcc.bnl.gov/users/yesw/singularity/alma9-x86/modern-linuxtools/setupMe.sh

# Then use copilot-api as normal
copilot-api auth
copilot-api start

💻 Usage Examples

# Use with Aider
export ANTHROPIC_BASE_URL=http://localhost:4141 && aider --no-git --anthropic-api-key dummy --model anthropic/claude-sonnet-4.5

# Or use with Claude Code CLI
export ANTHROPIC_BASE_URL=http://localhost:4141 ANTHROPIC_AUTH_TOKEN=dummy ANTHROPIC_MODEL=claude-sonnet-4.5 && claude-code

📌 Important Notes:

• Use your own URL in ANTHROPIC_BASE_URL and remove trailing /
• Enable X11 forwarding when SSH-ing: ssh -X username@hostname
• All GitHub Copilot models (excluding Market models) become accessible

🤖 Automated Setup with run-claude-copilot.sh

For a streamlined experience, this script automates the entire setup process for using Claude Code with GitHub Copilot models.

✨ Key Features:

Feature	Description
📦 Auto Dependency Management	Installs nvm, npm, copilot-api, and claude-code
⚡ Simplified Usage	Single command to start fully configured Claude session
🔄 Model Selection	Specify which Copilot model to use
🛠️ Utility Functions	Check usage, list models, update packages
🔗 Transparent Args	Forwards arguments directly to claude command

💻 Usage Examples:

# Run Claude with default settings
./scripts/run-claude-copilot.sh

# List available Copilot models
./scripts/run-claude-copilot.sh --list-models

# Check your Copilot API usage
./scripts/run-claude-copilot.sh --check-usage

# Run Claude with a specific model and pass a prompt
./scripts/run-claude-copilot.sh --model claude-sonnet-4 -- -p "Explain quantum computing"

# Get help on the script's options
./scripts/run-claude-copilot.sh --help

# Get help on Claude's own options
./scripts/run-claude-copilot.sh -- --help

---

📚 Detailed Tool Guides

Comprehensive documentation for each AI terminal tool:

Tool	Description	Guide
🤝 Aider	AI pair programming in your terminal	Read Guide
🤖 GitHub Copilot CLI	Copilot coding agent directly in your terminal	Read Guide
💎 Gemini CLI	Google's Gemini in your terminal	Read Guide
🚀 Qwen Code	Qwen 3.6 Plus models in your terminal	Read Guide

---

🖥️ AI-Enhanced Terminals

⚡ Warp Terminal

AI-first terminal that integrates intelligent agents directly into the command line.

✨ Key Features:

Feature	Description
💬 Natural Language Commands	Generate commands with # trigger
🤖 Real-time AI	Autosuggestions and error detection
🎤 Voice Commands	Multi-agent parallel workflows
🏢 Enterprise Ready	SAML SSO, BYOL, zero data retention

📊 Usage Limits:

• 🆓 Free tier: 150 requests/month
• 💎 Paid plans available for higher usage

📦 Installation:

brew install --cask warp    # macOS
winget install Warp.Warp    # Windows

# Linux - Multiple package formats available
# See: https://www.warp.dev/blog/warp-for-linux
# Packages include: .deb (apt), .rpm (yum/dnf/zypper), Snap, Flatpak, AppImage, and AUR

🌊 Wave Terminal

Open-source terminal that brings graphical capabilities into the command line.

✨ Key Features:

Feature	Description
🖼️ Inline Previews	Images, markdown, CSV, video files
📝 VSCode-like Editor	Integrated editor for remote files
🌐 Built-in Browser	Web browser and SSH connection manager
📊 Custom Widgets	Dashboard creation capabilities
🖥️ Cross-platform	Local data storage for privacy

🤖 AI Integration:

• ✅ Built-in AI assistance for command suggestions
• ⚙️ Configurable AI models via "Add AI preset..."
• 🦙 Support for Ollama and other local models
• 🎯 Context-aware recommendations

📦 Installation:

Download from waveterm.dev/download

Available as: Snap, AppImage, .deb, .rpm, and Windows installers

📟 iTerm2 AI

Native AI integration for macOS's most popular terminal emulator.

✨ Key Features:

• 🧠 Built-in AI Chat: Interact with LLMs directly within iTerm2 windows
• ✍️ Command Composer: Describe what you want to do in English, and it generates the shell command
• 🔍 Code Explanation: Highlight output or commands to get instant explanations
• 🔑 BYOK: Bring Your Own Key (OpenAI, Gemini, etc.) for privacy and control

📦 Setup:

1. Install iTerm2 (v3.5+)
2. Install the AI plugin (Settings > General > AI > Install)
3. Configure your provider (OpenAI, Gemini, etc.) and API key
4. Start using the AI features:
   • Cmd + Y: The Command Generator (The AI will generate the actual shell command but will not run it yet.)
   • Cmd + Shift + Y: The AI Chat
     • Use this for multi-turn conversations.
     • If you grant permissions, this chat can read your current terminal history and error messages to provide context-aware answers.

🧩 TmuxAI

AI-Powered, Non-Intrusive Terminal Assistant that works wherever tmux runs.

✨ Key Features:

• 🚀 Universal Compatibility: Works with any terminal emulator via tmux
• 👻 Non-Intrusive: Runs in a separate pane or window, keeping your workflow clean
• 🤖 Model Flexibility: Supports OpenAI and other compatible APIs
• ⌨️ Keyboard Centric: Designed for efficiency with tmux keybindings

📦 Installation: Prerequisite: tmux must be installed. Follow instructions at github.com/alvinunreal/tmuxai

---

Made with ❤️ by the Community

⭐ Star on GitHub | 🐛 Report Issues | 💡 Contribute

Supercharge your terminal workflow! 🚀