Rate Limits

Every chat platform rate-limits API calls. Bots must respect:

• Global limits: total calls per second.
• Per-route limits: e.g., 5 messages per channel per 5 seconds.
• Per-resource limits: 50 messages per minute per channel.

Discord:

• Returns 429 with retry_after header.
• x-ratelimit-remaining, x-ratelimit-reset headers.
• Per-bucket: routes share buckets.

Telegram:

• 30 messages per second to different users.
• 1 message per second per chat.
• 20 messages per minute per group.

Implementation:

Token bucket variant:

Defensive coding:

• Catch 429.
• Sleep retry_after.
• Retry once.
• Log if persists.

Spam prevention:

• Bots should not send many messages quickly.
• Even if API allows, users hate spam.
• Self-throttle: 1 message per 1-3 seconds for chatty actions.

Sharding (Discord):

• For very large bots (> 2500 servers), split into shards.
• Each shard handles subset of servers.
• Lower per-shard load.

Modern bot frameworks (discord.py, Bolt) handle rate limits automatically. For our toy: skip; focus on logic.