---
Status: Backlog
Owner Role: Developer
Created: 2026-05-22T07:40:00.000Z
Completed:
Depends On: 000300
Validation: node ai/runtime/dist/cli.js validate --tier chunk --chunk 000301 --json
---

# Runtime Health Minimum-Downtime Heavy Refresh Performance for cfd-0035

## Autopilot

- Repository/runtime state is authoritative; chat memory is not authoritative.
- Continue autonomously unless a stop condition is reached.
- Do not weaken freshness semantics to make slow data look fresh.

## Goal

Reduce or precisely diagnose slow Runtime Health refreshes and minimum-downtime issues.

## Scope

- Profile Runtime Health refresh path.
- Separate:
  - materialized read-model read latency
  - explicit refresh latency
  - audit/timeline boundedness cost
  - supervisor/service aggregation cost
  - filesystem scan cost
  - GraphQL/backend projection cost
- Avoid guessing N+1 unless measured.
- Add targeted optimizations only where measured.
- Keep stale/degraded freshness semantics truthful.
- Preserve Runtime/GraphQL as canonical UI truth.

## Acceptance Criteria

- Fresh read-model read remains fast.
- Explicit refresh bottlenecks are measured.
- Any optimization has before/after evidence.
- Remaining slow surfaces are precisely listed.
- `cfd-0035` is narrowed or a concrete follow-up is created.
- Existing validation passes.

## Validation Requirements

- Performance profile report.
- Runtime Health smoke.
- Backend health projection tests if touched.
- Summary validation.
- Canonical validation.

## Stop Conditions

- Stop if meaningful optimization requires broad Runtime Health rewrite.
- Stop if performance data is inconclusive.
- Stop if optimization would weaken freshness/confidence semantics.