Skip to content

getting-started.md

Latest commit

 

History

History
122 lines (89 loc) · 3.94 KB

getting-started.md

File metadata and controls

122 lines (89 loc) · 3.94 KB
Raw

Getting started

Antares is one process. It holds the agent, the HTTP API, the dashboard, the terminal interface, the scheduler, and the messaging bots. There is nothing else to install and nothing to orchestrate.

Requirements

  • Go 1.26 or newer to build from source.
  • Node or Bun to build the dashboard. Bun is what the Makefile reaches for first; npm works.
  • A model provider. Any OpenAI-compatible endpoint, or a native adapter for Anthropic, OpenAI, or Gemini. A local runtime like Ollama or LM Studio works too, with no key at all.
  • Chrome or Chromium if you want the browser tool. Everything else works without it.

Build and run

git clone https://github.com/enowdev/antares.git
cd antares
make install     # Go modules, Air, dashboard packages
make build       # one binary with the dashboard inside it
./bin/antares

The first run has nothing configured, so it drops straight into setup:

    1  Browser   — a guided page in the dashboard
    2  Terminal  — a few questions right here

Both write the same ~/.antares/config.yaml. Pick whichever is in front of you.

On a machine with no display, the browser option prints every address the setup page can be reached on, so you can finish from a laptop on the same network.

What setup asks

Which provider. Pick one and paste a key. Antares checks the key against the provider before saving it, so a typo fails there rather than mid-conversation. A local runtime needs no key — just confirm the address.

Which model. The list comes from the provider you just connected. Aggregators publish hundreds; the picker is searchable.

Where to keep data. SQLite by default, in ~/.antares/antares.db. Postgres if you would rather. Both schemas are created on first run.

How you want to talk to it. The terminal interface, the web dashboard, or both. This only sets the default; either is available at any time.

First conversation

./bin/antares

That is the terminal interface. Type a message and press Enter. Alt+Enter makes a new line, Ctrl+C interrupts a turn and quits on a second press.

Type / for the command palette — /help lists everything, /status shows what is configured, /model changes the model without leaving the conversation.

For the dashboard instead:

./bin/antares serve

Then open http://localhost:8787. The chat there is the same agent over the same sessions and the same memory; you can start a conversation in one and continue it in the other.

Reaching it from another machine

Antares binds 0.0.0.0 by default, so a private-network address works with no further setup:

http://<host-or-tailscale-ip>:8787

The dashboard is left open while server.auth_token is empty, which is the right default behind a private network and the wrong one anywhere else. To require a token:

antares config set server.auth_token "$(openssl rand -hex 24)"

Clients then send Authorization: Bearer <token>. The dashboard prompts for it and remembers it.

Before a bearer token exists, the setup wizard's mutating requests and the first dashboard-password operation accept only loopback connections. Use the local dashboard, an SSH tunnel, or configure the bearer token from the CLI before bootstrapping remotely.

Where things live

~/.antares/
  config.yaml       every setting
  .env              secrets, loaded automatically
  antares.db        sessions, messages, memory, vectors, schedules
  skills/           the skill library
  browser/          the browser profile, so logins survive a restart
  screenshots/      captures the browser tool wrote
  logs/             if logging.file is set

ANTARES_HOME moves all of it somewhere else, which is how you run two independent instances on one machine.

Next

  • Configuration — every setting and where it can be set.
  • Tools — what the agent can actually do.
  • Commands — the slash commands.