Installation
This guide will walk you through installing the Dexto CLI and setting up your environment so you can start running agents.
Prerequisites
- No runtime dependency is required for native install.
- Node.js >= 20.0.0 is only required if you build from source.
Optional: An LLM API Key (not required for local models)
- Get a Gemini Key (free option available)
- Get a Groq Key (free option available)
- Get an OpenAI Key
- Get a Claude Key
1. Install Dexto
Install Dexto using the native installer:
# macOS / Linux / WSL
curl -fsSL https://dexto.ai/install | bash
# Windows PowerShell
irm https://dexto.ai/install.ps1 | iex
This adds the dexto command to your system and writes install metadata to ~/.dexto/install.json.
Upgrade
# latest
dexto upgrade
# pinned
dexto upgrade 1.6.8
Uninstall CLI
# default: remove CLI binary only
dexto uninstall
# remove ~/.dexto too
dexto uninstall --purge
PATH Troubleshooting
If you had an older npm global install, dexto upgrade automatically migrates it to native.
If your shell still resolves the wrong binary:
which -a dexto
Remove stale entries and run dexto upgrade again.
2. Run Setup
dexto
This triggers the first-time setup wizard.
Quick Start (Recommended)
- Choose setup type → Select "Get started now"
- Pick provider → Gemini (free), Groq (fast), or Local
- Get API key → Browser opens, or paste existing key
- Confirm mode → Press Enter for Terminal (default)
Done! You're ready to chat.
Custom Setup
- Choose "Choose my own provider"
- Browse providers → OpenAI, Anthropic, local models, gateways
- Select model → Pick from available models
- Configure API key → Browser, paste, or skip
- Choose default mode → Terminal, Browser, or API Server
Supported Providers
| Category | Providers |
|---|---|
| Free Cloud | Google Gemini, Groq |
| Local (No API key) | Local Models, Ollama (requires Ollama installed) |
| Cloud | OpenAI, Anthropic, xAI, Cohere |
| Gateways | OpenRouter, Glama, LiteLLM, OpenAI-Compatible |
| Enterprise | Google Vertex AI, AWS Bedrock |
Modes Explained
| Mode | Flag | Description | Best For |
|---|---|---|---|
| Terminal | --mode cli | Interactive CLI in your terminal | Quick tasks, coding |
| Browser | --mode web | Web UI at localhost:3000 | Long conversations |
| API Server | --mode server | REST API on port 3001 | Integrations, apps |
3. Start Using Dexto
After setup, Dexto launches in your selected default mode.
# Run with your default mode
dexto
# Override with a specific mode
dexto --mode cli
dexto --mode web
dexto --mode server
# One-shot commands (auto-uses CLI mode)
dexto "say hello"
dexto -p "list files in this directory"
Reconfigure Anytime
# Open settings menu
dexto setup
# Force re-run full setup
dexto setup --force
Non-Interactive Setup
For automation or CI environments:
dexto setup --provider google --model gemini-2.5-pro
dexto setup --provider ollama --model llama3.2
dexto setup --quick-start
For detailed information about all CLI commands, flags, and advanced usage patterns, check out our comprehensive CLI Guide.
Next Step: Build Your First Agent
Now that Dexto is installed, you're ready to create your first custom agent with its own configuration and capabilities.
Continue to the Build Your First Agent Tutorial to learn how to build agents using declarative configuration.