Legion CLI

The AI coding agent built for the terminal. Generate code from natural language, automate tasks, and run terminal commands — powered by 500+ AI models.

Your AI pair programmer lives in your terminal now. No browser tabs. No switching apps. Just pure productivity.

🚀 Quick Demo

Click the button above to watch Legion turn a natural language request into working code in real-time.

⚡ Install

Get started in literally 30 seconds:

npm install -g @legioncli/cli

No setup. No config files. Just:

legion

Or skip the install entirely:

npx @legioncli/cli

🎮 Getting Started

Interactive Mode — Chat with Your Agent

legion

Spin up a beautiful TUI and have a real conversation with your AI partner. Type naturally. Get code. Done.

One-Shot Mode — Fire & Forget

legion run "add input validation to the signup form"

The agent reads your codebase, understands the context, runs the changes, and gets out of your way.

🔥 What Legion Actually Does

💻 Code Generation

Stop copy-pasting from Stack Overflow. Describe what you need in English. Legion writes it, tests it, and explains what it did. Works with any language, any framework.

⚡ Terminal Command Execution

Want the agent to run commands? Let it. Git commits, tests, deployments, environment setup — whatever. You see every command before it executes.

🧠 500+ AI Models on Tap

Not stuck with one LLM. Legion supports:

• OpenAI (GPT-4, o1, o3-mini, and more)
• Anthropic (Claude 3.5, Sonnet, Opus)
• Google (Gemini, all versions)
• Meta, Mistral, Groq, and 50+ others

Pick the right model for the job. Switch on a whim.

🔌 Model Context Protocol (MCP)

Extend Legion's brain with your own tools. Connect to APIs, databases, file systems, anything. The agent can now reach way beyond just code.

🎯 4 Agent Modes (+ Custom)

• Plan Mode — Think through complex problems step-by-step before coding
• Code Mode — Full speed code generation with zero overthinking
• Debug Mode — Feed it errors and broken code, watch it fix things
• Custom Mode — Build your own agent personality with your own system prompt

Pick the vibe that matches your workflow.

💾 Sessions & Persistent Memory

Everything is saved. Pick up a conversation from last week. Export transcripts as markdown or JSON. Your entire conversation history lives in Legion.

🧠 Adaptive Responses

Legion detects your mood. Frustrated? It slows down and explains more. Moving fast? It keeps pace. Excited? It matches your energy.

🛠️ Built-In Power Tools

• Test Runner — Run tests, understand failures, fix them
• Git PR Tools — Create PRs, manage branches, handle merges
• Memory System — Structured memory that persists across sessions
• And more — bash execution, file operations, codebase analysis

🎛️ The Model Picker (New)

We built something special for choosing your AI engine:

• Collapsible Providers — Expand/collapse OpenAI, Anthropic, Google, and 50+ others
• Full Pricing Breakdown — See input/output/reasoning/cache costs upfront before you pick
• Provider Stats — Know how many models each provider has
• Live Model Count — Total available models in your header (spoiler: it's 500+)

Pick the cheapest model for small tasks. Go for the smartest model when it matters.

🔄 Update & Maintenance (Auto-Magic)

# Check for updates and install with confirmation
legion update

# Skip the prompt and update instantly
legion update -y

# Or jump to a specific version
legion upgrade v2.5.0

Smart timeouts protect you on slow networks:

• 15s timeout for checking for new versions
• 120s timeout for the actual upgrade

No hanging. No frozen terminals.

💪 Real-World Examples

Generate a Full Component

legion run "create a React auth form with email/password validation, error handling, and loading states"

Debug Production Issues

legion run "we're getting 401s on the checkout endpoint — read the logs and fix it"

Automate Your Release

legion run "bump version to 2.1.0, update CHANGELOG, commit, create a git tag, and push"

Write Tests for Existing Code

legion run "write comprehensive unit tests for the payment processor function"

Refactor Legacy Code

legion run "this function is 200 lines and does 5 things — break it into smaller functions"

🎯 Why Legion Hits Different

✅ Zero Context Switching — Everything happens in your terminal. No browser tabs. No Slack notifications breaking your flow.

✅ Understands Your Codebase — Legion reads your files, runs your tests, interprets your errors. It's not guessing.

✅ Actually Executes — It doesn't just write code. It runs commands, creates commits, deploys. You see every step.

✅ You Control the AI — Bring your own API keys. Use any model. Switch providers whenever you want.

✅ Fully Transparent — Every command, every change, every decision is visible to you. You're never flying blind.

✅ Built for Speed — No bloat. No unnecessary confirmations. Fast enough to feel like an extension of your fingers.

📋 Command Reference

| Command | What It Does | |---------|--------------| | legion | Launch interactive agent mode | | legion run "<task>" | Execute a single task | | legion auth | Configure your API keys | | legion models | Browse the 500+ models (with pricing) | | legion mcp | Install and manage MCP servers | | legion session list | See all saved conversations | | legion session delete <id> | Remove a session | | legion export <id> | Export session as markdown/JSON | | legion update | Check for updates | | legion upgrade | Jump to a specific version |

Full list: legion --help

📄 License

MIT

Built for developers who move fast and hate context switching.

Keywords: AI coding agent, CLI tool, code generation, terminal, LLM, automation, developer productivity, MCP, multi-model AI