Soul-driven AI agent with permission-hardened tools, token budgets, and multi-channel access. Runs 24/7 from CLI, Telegram or More.
---
id: setup
title: First-Time Setup
---
Run `mercury` for the first time. The onboarding wizard asks for:
- **Your name** — used by the agent to address you
- **Agent name** — defaults to "Mercury", customize it
- **LLM providers** — choose one or more: DeepSeek, OpenAI, Anthropic, Grok, Ollama Cloud, Ollama Local, Atomic Chat, or OpenAI Compatible
- **Model selection** — after each valid key, Mercury fetches recommended chat models and lets you choose a default
- **Telegram bot token** — optional, for remote access in private 1:1 chats
Provider keys are validated by format and then by fetching available models. If model discovery fails, Mercury asks you to re-enter the key instead of accepting random text.
If you enable Telegram during first-time setup, Mercury starts a temporary pairing flow: send `/start` to your bot in Telegram, copy the pairing code you receive there, then paste that code into the CLI to finish setup.
All config is stored in `~/.mercury/` — nothing in your project directory.
:::info[Masked keys]
When reconfiguring with `mercury doctor`, existing API keys are shown masked like `sk-x••••4f2a`. Press Enter to keep the current value, or paste a new key to replace it.
:::