Spec 0237 — Refactor: stop losing cycle-start anchor events on the first `--push-to-main` resync by adding a `queue_state flush` subcommand and wiring `/dev-next` to call it at step 12
Frontmatter
Field
Value
kind
dev
spec
0237
slug
queue-state-flush-mechanism-for-cycle-start-anchor-events
title
Refactor: stop losing cycle-start anchor events on the first `--push-to-main` resync by adding a `queue_state flush` subcommand and wiring `/dev-next` to call it at step 12
type
refactoring
label
refactoring
version_bump
PATCH
target_version
TBD
status
deferred
depends_on
[]
complexity
S
created
2026-05-27
queued_at
started_at
merged_at
deployed_at
pr
handover
failure_step
source_session
promoted_from_draft
disposition
ship
disposition_reason
The cycle-start anchor events (cycle_started / preflight_ok / handoff_read / spec_read / planning_started / reconcile_complete) are silently lost every `/dev-next` cycle today, which means the dashboard's stage-duration breakdown for the pre-implementation phases is structurally broken — observability we explicitly added via spec 0156 is regressed and must be repaired.
Event timeline
When
Step
Data
2026-05-27T13:30:28Z
queued
{}
Links