Define cascade failure detection — downstream impact tracking

[koad]

Monitor and alert when a single entity's outage cascades to dependent services.

Pattern: Entity X publishes/deploys → downstream entities can't work:

• Vulcan deployment breaks → all entities waiting for builds
• Mercury publishing pipeline fails → Muse content doesn't ship
• Veritas gate down → no reviews approved, nothing moves
• Vesta infrastructure change → outage across dependent entities

What to monitor:

• Pipeline dependency graph: which entities depend on which for unblocking
• Time-series analysis: when one entity stalls, do others follow within N hours?
• Failure timing: unexpected stalls in Muse immediately after Mercury last commit
• Access anomalies: rapid spread of permission errors across multiple repos

Tier: CRITICAL if cascade affects >3 entities or core path (Juno, Vulcan, Veritas)

Action:

1. File CRITICAL alert to koad (Keybase) if cascade detected
2. File issue on primary entity (the root cause)
3. File info-level summary on dependent entities
4. Cross-ref all in Juno for orchestration

Related: alert-routing.md, fourty4-watch-setup.md

Enables root-cause diagnosis; prevents treating symptoms as independent problems.