---

name: cascade-and-evolution description: "Use this sub-skill for EverOS Markdown storage layout, cascade sync/rebuild/backfill, OME strategies, reflection, and operational recovery." disable-model-invocation: true metadata: disco-role: operating license: Apache 2.0

EverOS Cascade and Evolution

Use this sub-skill when a task asks about the memory-root layout, Markdown source of truth, SQLite/LanceDB indexes, everos cascade operations, OME strategy configuration, backfill, reflection, or operational recovery.

Read/run map

• Read storage and cascade for memory-root layout, Markdown/SQLite/LanceDB roles, cascade daemon loops, and CLI commands.
• Read evolution and reflection for OME strategy behavior, ome.toml, manual trigger, reflection, and backfill phases.
• Read troubleshooting for schema drift, queue failures, rebuild safety, file descriptor exhaustion, watcher issues, provider gates, and server-lock conditions.
• Run everos_status_probe.py for safe /health and/or everos cascade status probes. It does not run destructive operations.

Command safety

Command	Safe with running server?	Notes
everos cascade status	yes	Read-only queue/LSN summary.
everos cascade sync [PATH]	yes	Drains pending queue; can force-enqueue one Markdown path.
everos cascade fix	yes	Lists failed rows.
everos cascade fix --apply	generally yes	Requeues retryable failures and drains once.
everos cascade rebuild	no	Stop server first; it drops/recreates LanceDB business tables.
everos cascade backfill	phases 2/3 need exclusive OME access	Uses providers and can spend model/embedding budget.

Key model

EverOS writes Markdown as truth, records durable queue/state in SQLite, and derives LanceDB rows asynchronously. If derived indexes are wrong, rebuild from Markdown; do not treat LanceDB as the canonical store.