Skip to main content
This is a quick path to a safe, local setup for your first bot. This guide walks you through a local setup on a single machine (your “gateway host”). Required reading: Security and pairing (safe defaults, recommended setups, and risks).

​
Prerequisites

  • The moldable CLI installed and available in your PATH
  • Network access for your chosen channel and AI backend

​
1) Install the CLI

See Install, then verify: 
moldable --version

​
2) Run onboarding (guided)

moldable onboard
Pick a recommended setup (e.g. “Just me”, “Local network”, “Telegram”, “Webhooks (e.g. WhatsApp Cloud)”), and we’ll configure safe defaults for you.
Choose Manual (advanced) only if you know exactly what you’re doing. Safe defaults you get automatically: local‑only access, public access off, auth required, and pairing enabled. If you choose Webhooks, you’ll need a tunnel so the webhook endpoint is reachable without exposing the gateway. See Security and pairing.

​
3) Start the gateway

moldable gateway run
If the startup security summary reports critical findings, the gateway will refuse to start. Fix the issues or run with --force if you intentionally accept the risk.

​
4) Pair a user (DMs)

When a new user sends a DM, the gateway creates a pairing request. Approve it from the CLI: 
moldable gateway pair list
moldable gateway pair approve telegram <code>

​
5) Check status

moldable gateway health
moldable gateway status
Optional: run moldable gateway audit for a quick checklist-style safety scan.

​
Where files live

  • Config: ~/.moldable/gateway/config.json5 (JSON5 or JSON)
  • State: ~/.moldable/gateway/ (local gateway data)
You can override the config path with --config on any CLI command. Next: Configuration