Quick Start

This page gives both interactive and headless command paths for Telegram and Discord. For platform setup details, troubleshooting, and production examples, continue to the guides:

• Channel Configuration
• Provider Configuration
• MCP Integration
• Deployment (Docker and Compose)

Human-friendly interactive mode (default)

Run these commands without extra flags and follow the prompts:

llm-expose add model
llm-expose add channel
llm-expose add pair
llm-expose start

Use this path for first-time setup and manual operations.

Headless mode (automation)

For scripts and CI, use --no-input and include all required values. Add -y when confirmations can appear.

Telegram in 4 steps

1. Add a model:

llm-expose add model \
    --name gpt4o-mini \
    --provider openai \
    --model-id gpt-4o-mini \
    -y --no-input

1. Add a Telegram channel:

llm-expose add channel \
    --name team-telegram \
    --client-type telegram \
    --bot-token "123456789:AAExampleTelegramToken" \
    --model-name gpt4o-mini \
    -y --no-input

1. Pair the allowed Telegram chat ID:

llm-expose add pair 123456789 --channel team-telegram --no-input

1. Start the runtime:

llm-expose start --channel team-telegram -y --no-input

Discord in 4 steps

1. Reuse the same model or create a dedicated one:

llm-expose add model \
    --name gpt4o-discord \
    --provider openai \
    --model-id gpt-4o-mini \
    -y --no-input

1. Add a Discord channel config:

llm-expose add channel \
    --name ops-discord \
    --client-type discord \
    --bot-token "YOUR_DISCORD_BOT_TOKEN" \
    --model-name gpt4o-discord \
    -y --no-input

1. Pair the allowed Discord channel ID:

llm-expose add pair 987654321098765432 --channel ops-discord --no-input

1. Start the runtime:

llm-expose start --channel ops-discord -y --no-input

Add MCP servers (optional)

1. Add an MCP server definition:

llm-expose add mcp \
    --name web-search \
    --transport stdio \
    --command uvx \
    --args mcp-server-web-search \
    --tool-confirmation never \
    -y --no-input

1. Attach it to a channel:

llm-expose add channel \
    --name team-telegram \
    --client-type telegram \
    --bot-token "123456789:AAExampleTelegramToken" \
    --model-name gpt4o-mini \
    --mcp-server web-search \
    -y --no-input

See MCP Integration for transport-specific server configs and approval policies.

Verify configuration

llm-expose list models
llm-expose list channels
llm-expose list pairs
llm-expose list mcp

Use llm-expose <command> --help and llm-expose <command> <subcommand> --help for full options.