Waterbrother is built around local control. This page explains what the site and product store, what external services are involved, and what users should assume when enabling messaging integrations.
The public Waterbrother site is a static docs and product site. It does not require an account to browse. If analytics or hosting logs are enabled by the platform, they may include IP address, browser metadata, and page request information.
The CLI stores local configuration, session history, receipts, and shared-project state on the machine where Waterbrother runs. That can include prompts, file paths, command traces, runtime configuration, and collaboration state such as room membership, tasks, and invite ids.
When Telegram is enabled, Waterbrother stores Telegram user ids, usernames when available, chat ids, session routing, pairing state, and shared-room membership state needed for the gateway to function. Messaging providers also have their own privacy policies and infrastructure outside Waterbrother.
Waterbrother is bring-your-own-model. Prompts and context may be sent to the provider you configure, such as OpenAI, Anthropic, OpenRouter, Ollama, xAI, or another compatible API. Your use of those services is governed by their own policies and retention rules.
If you enable messaging, shared rooms, or external model providers, you are responsible for deciding what data is appropriate to process through those surfaces. For sensitive work, review your local config, gateway setup, room membership, and model provider settings before use.
