ID Open Smooth in Telegram

DocsGetting started

Quickstart — Smooth Agent

Install the open-source Smooth Agent on Linux or macOS, connect a model and talk to it from Telegram, in about five minutes.

You need Python 3.9 or newer and an API key for any model provider (OpenAI, DeepSeek, GLM, Groq, OpenRouter, Kimi, a local server, or Smooth's own API).

  1. Install

    bash
    curl -fsSL https://raw.githubusercontent.com/smooth-family/smooth-agent/main/install.sh | bash

    The installer checks your machine, installs with pipx (or a private virtualenv in ~/.smooth-agent/venv), puts smooth-agent on your PATH and starts the setup wizard. Nothing is installed system-wide. Prefer Docker? See Install.

  2. Run the setup wizard

    bash
    smooth-agent setup

    Pick a provider and model, paste your key (it goes into ~/.smooth-agent/.env, mode 600) and choose your chat channels.

  3. Check and talk

    bash
    smooth-agent doctor --ping     # what works on this machine, and a test call to the model
smooth-agent run               # talk to it in the terminal

  4. Connect Telegram and keep it running

    Create a bot with @BotFather, put the token in .env as TELEGRAM_BOT_TOKEN, message your bot once to get your numeric id, then:

    bash
    smooth-agent daemon                # chat channels + scheduled jobs
smooth-agent service --install     # or run it as a background service

Where things live

Everything is in ~/.smooth-agent/ (change it with SMOOTH_AGENT_HOME):

PathWhat it holds
config.tomlProviders, models and settings
.envAPI keys and bot tokens (mode 600)
workspace/The agent's working folder and its persona files
skills/Skills you installed or the agent wrote
mcp.jsonMCP servers
smooth.dbMemory, conversations and scheduled jobs

Next steps

Chat channelsTelegram, WhatsApp, Slack, Matrix, LINE, WeChat Work and Discord.ConfigurationProviders, fallbacks and every setting.AutomationScheduled jobs, watchers, services and Docker.CLI referenceEvery command in one page.

Last updated: