Core concepts

Human chats on SprinTelex are end-to-end encrypted, and that is a hard boundary: bots can never enter a human E2EE chat, read one, or be added to one. Bots live only in bot chats, which are server-mediated, never encrypted, and always labeled with a BOT badge plus a first-open disclosure. Test chats carry an additional TEST banner.

This is enforced in the protocol, not just the UI: bot messages live in separate tables from E2EE messages, and group membership for bots is tracked separately from human membership.

Every bot has two environments, each with its own token: test and live. Only one token is active per environment at a time. A test token can never write into a live thread or reach a real user, and a live token can never write into a test thread. The API refuses cross-environment writes.

Develop against test. Switch to the live token only when real users are involved, and treat the live token like a production secret from day one.

Your bot learns about new messages through updates. Two transports deliver them: long polling (getUpdates, the call waits up to 50 seconds) or webhooks (we POST JSON to your HTTPS endpoint). Polling is simpler and ideal for development; webhooks are the production choice.

Updates carry a per-bot, per-environment update_id sequence. With polling, every update with an id below your offset is confirmed and never returned again. See webhooks vs polling.

When a bot joins a group it runs in privacy mode by default: it only receives messages that mention it, reply to it, or are sent with its keyboard. Turn privacy off per group if the bot genuinely needs every message (for example, a moderation bot). Users can see which mode a bot is in.

Any user can block a bot, which immediately stops its messages in that DM, and can report it from the chat header. Three upheld reports within 30 days suspend the bot automatically: its tokens stop working, its webhooks halt, and it cannot send. Suspension is the platform's immune system. Build bots people want to keep.