Configuration

All configuration lives in environment variables, validated by a Zod schema in apps/discord/src/utils/envSchema.ts. Invalid config exits the process immediately — a missing or malformed value fails fast instead of misbehaving later.

Locally these come from apps/discord/.env. In production, inject them through your platform (see Deployment).

Required

Variable	Notes
`DATABASE_URL`	PostgreSQL connection string
`REDIS_URL`	Redis connection string
`DISCORD_APPLICATION_BOT_TOKEN`	Bot token
`OWNER_ID`	Snowflake of the bot owner
`MAIN_CHANNEL_ID`	Announcement channel snowflake

Optional (with defaults)

Variable	Default	Notes
`DATABASE_POOL_MAX`	`10`	Max Postgres pool size per process (1–100)
`DATABASE_QUERY_LOG`	`false`	Logs SQL query text (never parameters) at debug level. Development only
`WORKER_CONCURRENCY`	`4`	BullMQ worker concurrency per shard
`DISCORD_DEFAULT_STATUS`	`Spreading Paw-sitivity 🐾`	Fallback presence text
`DISCORD_DEFAULT_ACTIVITY_TYPE`	`Custom`	`Playing`, `Streaming`, `Listening`, `Custom`
`DEFAULT_ACTIVITY_URL`	—	Stream URL (for `Streaming`)
`DISCORD_ACTIVITY_INTERVAL_MINUTES`	`15`	Presence rotation interval (1–1440)
`ALLOWED_USERS`	—	Comma-separated snowflakes allowed to run `/admin`. Empty disables every `/admin` command
`HOST`	— (all interfaces)	Health API bind address. Leave unset in Docker/Dokploy; `127.0.0.1` binds loopback only
`PORT`	`3000`	Health API port, a number from 1 to 65535
`VERSION`	`apps/discord/package.json` version	Shown in `/about` and `/changelog`
`NODE_ENV`	`development`	`development`, `production`, `test`
`PREMIUM_ENABLED`	`false`	Premium gating. `false` opens custom schedules to every server (see below)
`DISCORD_PREMIUM_SKU_ID`	—	Required only when `PREMIUM_ENABLED=true`; an empty value counts as unset

SQL is not logged by default, including in development. Set DATABASE_QUERY_LOG=true to see query text while debugging.

What `PREMIUM_ENABLED=false` means

false turns Premium gating off; it does not hide custom scheduling. /setup schedule is open to every server, saved custom schedules apply, and there is no purchase button or entitlement handling. That suits a self-hosted bot. A bot that sells Premium must set PREMIUM_ENABLED=true and DISCORD_PREMIUM_SKU_ID; see Enable Premium.

Container-only settings

SKIP_MIGRATIONS=true is read by the Docker entrypoint, not by the bot, and skips the startup migration step. Use it only while you baseline or reconcile an existing database; with no public."Guild" table the bot exits with "Database schema missing" instead of starting; see Database & schema for how migrations run.

Postgres connection budget

DATABASE_POOL_MAX applies per process. The supervisor process and every shard process each open their own pool (connections are opened lazily and closed after 30 seconds idle; the supervisor only needs one or two). Keep:

shards × DATABASE_POOL_MAX + 2  ≤  max_connections − reserved connections

Discord's recommended shard count grows with the number of servers, so lower DATABASE_POOL_MAX (or add a pooler such as PgBouncer) as shards are added. Run PgBouncer in session mode: transaction mode needs prepared statements turned off in the client, which the bot does not do, and the migration step holds a session-level advisory lock.

Optional and unused

The bot does not read these. They are accepted (and validated as snowflakes when set) so older .env files keep working, and can be removed.

Variable	Notes
`DISCORD_APPLICATION_ID`	The bot uses `client.application` instead
`DISCORD_APPLICATION_PUBLIC_KEY`	Only needed for HTTP interactions, which the bot does not use
`MAIN_GUILD_ID`	Announcements use `MAIN_CHANNEL_ID` alone

If PREMIUM_ENABLED=true, DISCORD_PREMIUM_SKU_ID must be set — the schema rejects the config otherwise.

The current template always lives in apps/discord/.env.example.