Skip to main content

Quickstart

The fastest path from nothing to a working chat with Anyy. Five minutes, one command each step. This page stays deliberately short and links out to the detailed pages for anything you want to go deeper on.

Who This Is For​

You want to try Anyy right now on macOS or Linux and have a working conversation before reading anything else. You are comfortable running a couple of shell commands. You do not need to understand the architecture, set up a background service, or wire up messaging channels to get started — those come later, in Add the Next Layer.

If you would rather understand each piece first, start with Installation and Setup instead.

The Fastest Path​

Four commands. The first three get you to a working chat; the last confirms it.

curl -fsSL https://anyy.ai/install.sh | sh   # 1. install
anyy setup                                            # 2. configure a provider + model
anyy                                                  # 3. start chatting (TUI)
anyy status                                           # verify (in another terminal)

Running anyy with no arguments opens the terminal app and starts an embedded gateway for you — there is no separate "start the server" step on the fast path. The sections below walk through each command.

1. Install Anyy​

Anyy ships as a single native Go binary named anyy.

Install from a release artifact​

The canonical one-line installer downloads a checksum-verified release for your OS and architecture and installs it under your home directory. It does not need root and does not edit your shell startup files:

curl -fsSL https://anyy.ai/install.sh | sh

Confirm it landed on your PATH:

anyy --version
Anyy (anyy) version v0.1.57 channel stable commit 8627a748340f built 2026-06-15T08:03:45Z

If the command is not found, add the entrypoint directory (typically ~/.local/bin) to your PATH. For the full walkthrough, options, and the home directory layout, see Installation.

Build from source​

If you have a Go 1.26+ toolchain, build it yourself:

make build

This produces dist/anyy. See the source build section for how to promote a local build into the resident layout.

2. Run Initial Setup​

The first time, run the interactive wizard to choose a provider, authenticate, and pick a default model:

anyy setup

The wizard walks you through provider selection, default model, optional fallback routing, optional capabilities, and messaging channels. You can rerun any part later (anyy setup model, anyy setup channels, …). For the full reference, see Setup.

note

Anyy is provider-neutral. Providers and models are configurable backends you select here — Anyy itself always identifies as Anyy.

Choose a provider​

Pick your provider and complete its authentication in the wizard. To reconfigure only this later:

anyy setup model

Configure the default model​

The same anyy setup model step sets the default model used for new sessions. To add automatic fallback routing across models, run anyy setup model fallback. Details: Model Setup.

How settings are stored​

Setup writes a config.yaml under Anyy's home directory (default ~/.anyy), and stores credentials separately under secrets/. Nothing is written to your shell profile. See How settings are stored for the full layout.

3. Start Your First Chat​

There are two ways to talk to Anyy. For your first chat, use the TUI — it is self-contained.

Start the TUI​

anyy

With no arguments, anyy launches the interactive terminal app. If no gateway is already running, the TUI starts an embedded one automatically, so this single command is all you need. Type a message, press Enter, and watch the reply stream in. Full guide: TUI.

Use one-shot chat​

To send a single prompt and get the answer back in your shell — handy for scripting — use anyy chat:

anyy chat "Summarize what you can help me with."

Unlike the TUI, one-shot chat requires a running gateway. If you see Could not create chat session … dial unix … no such file or directory, start the gateway first:

anyy gateway start
anyy chat "Hello"

Add --json for machine-readable output. See Chat for roles, sessions, approvals, and exit codes.

4. Verify Sessions Work​

In a second terminal, confirm the runtime is healthy:

anyy status
◆ Anyy status
  Gateway             running
  Profile             default
  Default role        default
  Home                /Users/you/.anyy
  Timezone            Asia/Shanghai (host)

Gateway running means your chat is being served by a live gateway. For a deeper health check of providers, storage, and capabilities, run:

anyy doctor

Every message you send belongs to a session, and sessions persist across restarts and across surfaces (TUI, one-shot, messaging). To prove it, exit the TUI and resume where you left off:

anyy --resume SESSION_ID

More on continuity, naming, and search: Sessions.

5. Try Key Features​

A few things worth trying in your first session.

Use the terminal​

Ask Anyy to do real work and approve actions as they come up:

List the files in my current directory and tell me which are largest.

Anyy surfaces an approval card before state-changing or higher-risk operations, so you stay in control without approving ordinary read-only work. See CLI Interface for how the command-line entry point is organized.

Slash commands​

Inside the TUI, type / to open the command menu — list or resume sessions, start a new one, switch model, inspect status, compress long context, and more. See TUI slash commands.

Multi-line input​

Paste or compose multi-line prompts directly in the TUI input. See the TUI guide for the input keybindings.

Interrupt the agent​

If a reply is going the wrong way, press Esc to interrupt and take back the prompt — no need to wait for it to finish.

6. Add the Next Layer​

Once chatting works, grow Anyy into a resident assistant.

Run it resident​

Keep Anyy running in the background so it is always available — and so one-shot chat and messaging channels work without an open TUI:

anyy gateway start

For installing it as a real service (systemd, launchd) with autostart, see Running Resident.

Messaging channels​

Reach Anyy from chat apps instead of the terminal:

anyy setup channels

See Messaging for supported platforms and account setup.

Automation and tools​

Schedule recurring work with cron jobs:

anyy cron list

Skills​

Skills package reusable capabilities. List and install them:

anyy skills list

Details: Skills.

MCP servers​

Connect external tools via the Model Context Protocol:

anyy mcp status

Details: MCP Integration.

Dashboard​

Open a local web view of sessions, jobs, and runtime state:

anyy dashboard

Details: Dashboard.

Common Failure Modes​

SymptomLikely causeFix
anyy: command not foundEntrypoint dir not on PATHAdd ~/.local/bin to PATH, reopen shell (details)
anyy chat → dial unix … no such file or directoryNo gateway runninganyy gateway start, then retry (the TUI starts one for you)
Chat replies with an auth or model errorProvider not configuredRerun anyy setup model
anyy status shows Gateway stoppedGateway not startedanyy gateway start, or just launch anyy
Setup can't reach the providerNetwork or credentialsanyy doctor to pinpoint the failing subsystem

Recovery Toolkit​

When something looks wrong, these read-only commands tell you what state you are in:

anyy status            # one-line summary of gateway, profile, subsystems
anyy doctor            # deep health check of providers, storage, channels
anyy gateway status    # is the resident gateway alive?
anyy --version         # confirm which build is on PATH

To reconfigure without reinstalling, rerun the relevant setup step (anyy setup model, anyy setup channels). For a structured list of problems and fixes, see Troubleshooting.

Quick Reference​

GoalCommand
Installcurl -fsSL https://anyy.ai/install.sh | sh
Configure provider + modelanyy setup
Start chatting (TUI)anyy
One-shot promptanyy chat "…"
Check statusanyy status
Health checkanyy doctor
Start resident gatewayanyy gateway start
Resume a sessionanyy --resume SESSION_ID
Full usageanyy --help

Next Steps​