Checks OpenAI status on a configurable interval, default 1m
Notifies subscribers about new incident updates
Notifies subscribers when component status changes
Supports incident-only, component-only, and component-filtered subscriptions
Uses Redis for subscribers, delivery state, component checkpoints, and seen incident update versions
Supports Telegram supergroup topics via message_thread_id
Clears an existing Telegram webhook before long polling, for migration from webhook deployments
Includes Docker Compose for local Redis + bot runtime

Bot Commands

Command	Description
`/start`	Subscribe current chat or topic
`/stop`	Unsubscribe current chat or topic
`/status [component]`	Show current OpenAI status, optionally for one component
`/components`	Show all OpenAI component statuses
`/subscribe <incident	component
`/subscribe component <name	id
`/history [count]`	Show recent incidents, default 5, max 10
`/uptime`	Show component health overview
`/info`	Show chat ID and subscription settings
`/help`	Show command help

Quick Start

cp .env.example .env
# edit .env and set TELEGRAM_BOT_TOKEN
docker compose up --build

For local development without Docker:

go mod tidy
go run ./cmd/openai-status-bot

Configuration

Variable	Default	Description
`TELEGRAM_BOT_TOKEN`	required	Telegram bot token from BotFather
`REDIS_ADDR`	`localhost:6379`	Redis address
`REDIS_PASSWORD`	empty	Redis password
`REDIS_DB`	`0`	Redis database number
`OPENAI_STATUS_BASE_URL`	`https://status.openai.com`	OpenAI status page base URL
`POLL_INTERVAL`	`1m`	Status check interval
`HTTP_TIMEOUT`	`10s`	HTTP request timeout
`LOG_LEVEL`	`info`	`debug`, `info`, `warn`, or `error`

Notes

The first successful poll seeds Redis and does not send historical incidents. Notifications start from later changes.

Incident update dedupe tracks the update content/version, so edited Statuspage updates can notify again. Delivery is checkpointed after successful fan-out; retryable Telegram failures may be retried on a later poll without advancing the global checkpoint.