For group admins

Building v2

Core moderation

Per-event bot skins Join captcha List available commands Media restriction for new members Privacy policy Reload cached admins and settings Group rules and /newrules Language selection Anti sticker spam Welcome message and /newwelcome Identify music in voice notes

Utility

Health check from chat Admin configuration menu Block listed users from joining Today's birthdays Call the admins Rewrite social links Ping every member Upcoming birthdays YouTube search Delete scheduled reposts Cross-group post forwarding Receive forwarded posts Message analysis (reply_to_message dump)Giveaways Image search (v1's catch-all)Owner-only operations Reverse image search Voice transcription

Fun

Battle poll Complaint bit Random cause of death Roll an n-sided die Firecracker sequence Meme generator Partnered convention posters Random media from the group Ship two members Age guess (agify.io)Conversational AI replies Per-group custom commands Media distortion Drawing idea prompt Fortune cookie Gender guess (genderize.io)Sticker DB auto-reply Unearth a random old message

Platform

Citus schema, colocation, UUIDv7 Compiled hot path with benchmark gate Multi-provider LLM with configurable models OTel traces, Prometheus, structlog Mock Telegram + executable Gherkin Self-hosted Telegram Bot API server Workspace, 3 services, CI Blob storage: GCS, S3, local, memory Admin resolution and cache Group config repository with shared cache String catalog ported from v1 locales Multi-tenant registry and schema Mongo -> Citus backfill Fleet-wide analytics and the group directory, owners only Per-group analytics endpoints Per-group audit trail Group settings over HTTP, admin-gated OAuth2 token issuer for the Telegram Mini App Telegram-login JWT for the web console

Architecture Mini App API Development Testing the API Sandbox End-to-end suite Deploy Cutover Cutover bucket export Multi-tenant v1 feature map Scenario coverage audit

Features

Multi-provider LLM with configurable models

anthropic + openai-compatible; per-task routing; cost metering

DoneM0coreplatform

v1 scenarios: —
Ported: —
Green: —
Failing: —

anthropic + openai-compatible; per-task routing; cost metering

What it does

Behaviour that must not change

How to verify it

Compiled hot path with benchmark gate

cooldowns 1.9x compiled; dedupe/textmatch/captcha ship pure per the gate

OTel traces, Prometheus, structlog

trace_id stored on message_events

On this page

What it does Behaviour that must not change How to verify it