Quickstart

Use the macOS desktop app when you can. Use CLI + Dashboard when you are on Linux, cloud, SSH, or a terminal-first macOS setup.

Elephant Agent's product position is agency-first: it should help you do more without thinking less. The Personal Model is how the product keeps judgment, evidence, questions, and growth close to the person.

Path overview

macOS desktop app

Step	Action	Done when...
Download	Get the latest build from GitHub Releases.	Elephant Agent.app is installed locally.
Create	Open the app and create your first elephant.	The app has a local elephant to return to.
Configure	Choose provider, model, embedding posture, and curiosity effort.	Chat / Wake and Personal Model are ready.
Continue	Open Chat / Wake.	You are back in the same conversation path.

The desktop app is the primary surface for Chat / Wake, Personal Model, providers, skills, tools, herd, messaging, reminders, usage, and settings.

CLI + Dashboard

Step	Command	Done when...
Install	curl -fsSL https://elephant.agentic-in.ai/install.sh | bash	elephant is on your PATH.
Initialize	elephant init	Your first elephant and provider posture exist.
Check readiness	elephant status	Provider and embedding readiness are clear.
Continue	elephant wake	You are in the durable chat surface.
Inspect	elephant dashboard	You can inspect Personal Model, questions, evidence, and runtime state.

tip

After the first CLI setup, elephant wake is the normal daily terminal entrypoint.

1. Run init

elephant init

init is the first-use setup flow. It:

• binds the local identity
• lets you choose a personality preset
• grows the first elephant with a clean Personal Model
• captures the active provider configuration before real conversations begin
• prepares the first named elephant so wake is the natural next step
• surfaces the IM handoff so elephant gateway setup can open the IM chooser without leaving the main setup flow

2. Confirm readiness

elephant status

Use status before the first durable conversation. It tells you whether the active provider and local runtime posture are ready to go.

3. Enter wake

elephant wake

wake is the main conversational surface. That is where continuity stays alive and your elephant keeps learning.

For remote or cloud use:

elephant dashboard --no-open

That prints the local Dashboard URL without trying to open a browser on the remote machine.

Optional next steps

If you want to...	Run	Read
Inspect what Elephant Agent understands	elephant dashboard	Dashboard
Wire a messaging app	elephant gateway setup	Messaging
Install or inspect skills	elephant skills or /skills	Skills
Choose a provider or model	elephant provider or /providers	Providers and models

Wire IM

elephant gateway setup

Use this entrypoint when you want to choose an IM surface, configure it, or inspect IM readiness without remembering the provider-specific setup commands.

4. Create another elephant only when you need one

elephant herd new nova

Each elephant is an isolated Elephant Agent individual with its own continuity line, personality drift, and future wake history.

warning

Do not create a new elephant for every task. Create one when you want a separate continuity line.

Optional: non-interactive setup

For scripted setup, you can preconfigure the provider path directly:

elephant init --non-interactive \
  --elephant-name nova \
  --provider-id openai-compatible

For deeper provider setup, continue with Providers and models.