state-hub/workplans/STATE-WP-0090-projection-task-field-convergence.md

id	type	title	domain	repo	status	owner	topic_slug	created	updated	related	state_hub_workstream_id
STATE-WP-0090	workplan	Preserve task descriptions and explicit empty notes in Forge projection	infotech	state-hub	active	codex	infotech	2026-09-10	2026-09-10	STATE-WP-0086 RMGR-WP-0017 HFACT-WP-0001	894448a6-66ae-5215-b596-0f4a8bc2c509

HFACT-WP-0001-T02 repeatedly repairs six of 24 selected task projections after normal sync. Forge derivation truncates nested evidence and converts explicit empty intervention notes into null, disagreeing with file-authoritative deep consistency. Fix the original parser rather than applying another batch patch.

Reproduce and correct task-field derivation

id: STATE-WP-0090-T01
status: done
priority: high
state_hub_task_id: "47ecc1ce-ca64-567b-8253-4e2bf72b8aa4"

Retain nested task sections, stop at the next peer/ancestor heading or task block, match source-description cleanup and preserve absent/null/empty metadata meaning. Keep task identity, status, collision and retirement rules unchanged. Test parity with the existing consistency parser and real PostgreSQL/API convergence.

Promote the verified correction and prove the selected chain

id: STATE-WP-0090-T02
status: progress
priority: high
state_hub_task_id: "da0089fc-2c6f-52d7-91e9-83ee338caf2a"

Run relevant projection/consistency/API regressions, commit/push, consume the immutable CI image and prepare the existing Helm release with reused values and atomic rollback. Verify the exact promoted image and healthy primary, then sync the three selected repositories using RMGR-WP-0017 and read back all 24 task fields with zero manual task PATCH repairs. Preserve unrelated untracked evidence.

Retain evidence and close the bounded defect

id: STATE-WP-0090-T03
status: todo
priority: medium
state_hub_task_id: "ac047cb8-fc56-5a2e-8a82-c5179f8cde8b"

Record source/image/Helm revisions, test results and before/after convergence. HFACT T02 retains other selected-backlog readiness work; native admission and factory execution remain their existing owner tasks. Do not mark factory G1/G2 complete or introduce another synchronization service.

Validation: all five new parser cases fail on the previous derivation. The corrected projection/consistency/API suite passes 242 tests on a disposable PostgreSQL database. Actual selected-source comparison reproduces six drifted tasks before the fix and zero afterward across 24 tasks. No production data was used by the tests; live promotion and readback remain T02.

Promotion prepared: Forge commit 0747b4d95b passed CI. Immutable image main-0747b4d has registry digest sha256:78828c9e85848538878a27d2e3e6bebfd0cc7192c0961c92c749b27acca7b6f5. Offline container verification matches the tested parser source and passes nested section/empty-note cases. Server-side Helm preview permits only image, version labels and their rendered checksum; configuration and credential references are unchanged, and there are no migration changes. Previous Helm revision: 60.

Rollout revision 61 could not schedule the API surge pod: the 4,000m node had 3,935m requested, leaving 65m against the required 100m. Atomic rollback restored healthy old-image services at revision 62. Available Prometheus history contains 1,078 five-minute samples within the seven-day query window (partial coverage): MCP p99 2.35m, peak 30.45m. The reviewed retry lowers only the MCP CPU request from 50m to 10m, preserving one replica, its 500m CPU limit and memory settings. This reduces its guaranteed share under contention; bursts remain permitted. Nominal headroom becomes 105m, sufficient for this 100m API surge but too narrow to count as factory capacity admission. Broader capacity planning needs a live residual before closeout. No unrelated workload allocation is changed.