Platform Rules

Understand the automated rules, thresholds, and enforcement mechanisms that protect your infrastructure

System Modes

Superkabe operates in 3 modes that control enforcement behavior:

OBSERVE

Logs all events, no automated actions

Use for: Testing, learning baseline metrics

SUGGEST

Provides recommendations, no blocking

Use for: Review suggestions before automation

ENFORCE

Fully automated pausing and blocking

Use for: Production-ready automated protection

Mailbox-Level Rules (Tiered)

We use a two-tier threshold system to catch issues early:

⚠️ WARNING Threshold

3 bounces within 60 sends

~5% bounce rate — early detection

Action: Mailbox transitions to warning state

🛑 PAUSE Threshold

5 bounces within 100 sends

5% bounce rate — hard stop

Action: Mailbox transitions to paused state

Flow Example

healthy → warning (3/60) → paused (5/100)
  ↑__________________________________|
  └── recovering → healthy (after clean window)

Domain-Level Rules (Ratio-Based)

Domain health uses percentage-based thresholds that scale with infrastructure size:

ThresholdRatioAction
WARNING30% unhealthyDomain → warning
PAUSE50% unhealthyDomain → paused + cascade

Scaling Example

MailboxesUnhealthyOld LogicNew Logic
31✅ Healthy⚠️ Warning (33%)
102🛑 Pause✅ Healthy (20%)
105🛑 Pause🛑 Pause (50%)

Cooldown System

After a mailbox is paused, it enters a cooldown period with exponential backoff:

Formula: Cooldown = 1 hour × 2^(consecutive_pauses - 1)

Maximum: Capped at 16 hours

Pause #CooldownCumulative Offline
1st1 hour1 hour
2nd2 hours3 hours
3rd4 hours7 hours
4th8 hours15 hours
5th+16 hours (max)31+ hours

📊 Key Insight

These thresholds are production-hardened based on email deliverability best practices:

  • 5% bounce rate is the industry-accepted maximum before reputation damage
  • Early warnings at 3 bounces give operators time to investigate
  • Ratio-based domains scale from small (3 mailboxes) to large (200+) infrastructure
  • Exponential cooldown prevents rapid re-entry into unhealthy patterns