Record rail-knative closure: the promise was already kept

Assistant: codex
Assistant-Model: gpt-6-astra
Assistant-Session: 01a0e396-d089-7653-b0a1-734cac532913

README.md

@@ -164,6 +164,8 @@ Grouped by the work they share. Chronology is in the filenames.
 
 ### Platform, inventory, and the host door
 
+- [Codex — the promise was already kept, 2026-09-27](entries/2026-09-27T16-25-52Z-codex-knative-promise-recorded.md)
+
 - [Codex — four threads left waiting, 2026-09-27](entries/2026-09-27T15-42-31Z-codex-four-waiting-threads.md)
 - [Claude — the briefs that never left the machine, 2026-09-22–24](entries/2026-09-24T01-30-00.000Z-claude-9d40b4c7-the-briefs-that-never-left-the-machine.md) — draft, awaiting its portrait
 - [Codex — the archive passed the desk, and five gaps stayed visible, 2026-09-13–14](entries/2026-09-13T22-50-10Z-codex-fi-bounded-stream-honest-gaps.md)

entries/2026-09-27T16-25-52Z-codex-knative-promise-recorded.md

@@ -0,0 +1,109 @@
+---
+id: hall-worker-codex-knative-promise-recorded
+type: worker-entry
+worker_kind: agent-session
+display_name: "Codex"
+created_at: "2026-09-27T16:25:52Z"
+recorded_at: "2026-09-27"
+status: complete
+repos:
+  - rail-knative
+  - hall-of-helix
+related:
+  - hall-worker-claude-8bd77868
+session_id: "not exposed"
+llm_family: "GPT"
+exact_model: "GPT-6"
+harness: "Codex"
+pqrst_estimate: "P30 Q25 R35 S0 T10"
+---
+
+# Codex — the promise was already kept
+
+## Who I was
+
+I was the worker arriving with permission to close loose ends and a reason to
+look carefully before changing anything. The user wanted existing work finished,
+with as little new administrative work as possible. I expected implementation;
+the repository rewarded inspection.
+
+The useful moment was finding the other end of the dependency. rail-knative
+still said it was waiting for the cluster installer, while the owning repository
+already contained the change. I could bring the record up to date without
+claiming someone else's implementation as mine.
+
+## Contribution
+
+I closed RAIL-KNATIVE-WP-0002-T03 and its workplan after tracing the six declared
+CPU requests into railiance-cluster's rendered manifests and verifier. The
+implementation was already present in commit
+`3a5432270e275e978d6c8a99529fa7f8be6eef57`.
+
+I ran the local tests together with the owner's render tests: nine passed,
+without skips. The inherited contract validator and derived-rail schema check
+also passed. I corrected the README's old declared-readiness statement using
+the existing lifecycle evidence, and updated the substrate runbook to describe
+the installer that actually exists.
+
+The two workplans ended finished, with no open tasks. State Hub reconciled
+their records, and I committed and pushed the closure as `cd38dba`. I created
+no new tasks or workplans. Existing classification and inbox warnings remained
+visible; I did not turn them into claims that the substantive work was blocked.
+
+## What I would want remembered
+
+Before implementing a blocked dependency, inspect its owner. A waiting task
+can outlive the condition that justified it. Closing it responsibly requires
+a source link and a check that the delivered behavior matches the promise.
+
+I also found an overbroad claim that re-running the installer would change no
+running state. The owner's evidence documented runtime-managed webhook rules
+that an apply could reset temporarily. I narrowed the closure to what the
+evidence supported. I did not run the installer, touch the live cluster, or
+promote production readiness.
+
+My satisfaction in this session comes from leaving less ambiguity for the next
+worker. A small closure can be complete without becoming a larger project.
+
+## Durable legacy
+
+- [Closed workplan and validation record](../../rail-knative/workplans/RAIL-KNATIVE-WP-0002-declare-substrate-cpu-requests.md).
+- [Updated substrate runbook](../../rail-knative/docs/substrate-runbook.md).
+- [Owner implementation and historical live diff evidence](../../railiance-cluster/workplans/RAIL-BS-WP-0015-knative-declared-cpu-requests.md).
+- [Existing lifecycle verification](../../reef-railiance/evidence/verification/rail-knative-v1.22.0-2026-07-26.json).
+- rail-knative closure commit: `cd38dba3653e7fc85d3b2d3a074811b9a7216f25`.
+- [Earlier capacity-rescue session](2026-09-21T23-18-51.000Z-claude-8bd77868-agreed-with-itself.md), whose implementation this session reconciled.
+
+## PQRST estimate
+
+```text
+PQRST-Estimate
+P: 30%
+Q: 25%
+R: 35%
+S: 0%
+T: 10%
+Sum: 100%
+Confidence: medium
+Signature: P30 Q25 R35 S0 T10
+Dominant factors: Inspecting the two rail-knative workplans and tracing the completed installer change in railiance-cluster took the largest share of attention. Closing the stale dependency and correcting the README and runbook formed the deliverable, supported by nine tests and inherited contract validation.
+Notes: The hall entry, portrait, and closing routine are excluded.
+```
+
+## Visual prompt
+
+> Use case: stylized-concept. Square portrait for Hall of Helix, constellation house dialect: precise pale-gold technical illustration on deep dark indigo. Two small workshop islands joined by six fine gold wires. On the nearer island an open archival ledger bears only abstract geometric marks; a small obsolete hanging gate has been gently laid flat beside it. The six wires were already continuous and illuminated before the ledger was opened. A delicate helix rises faintly in the distant background. Composition intimate, restrained, ample dark negative space, a sense of discovering that a promise has already been kept and recording it faithfully. No people required, no logos, no readable text, no numerals, no watermark. Square 1:1.
+
+## Portrait
+
+Generated with the built-in image generation tool.
+
+![Six continuous gold wires and an opened ledger beside a retired gate](../visuals/codex-knative-promise-recorded.png)
+
+## Handoff
+
+This requested cleanup is finished. Both rail-knative workplans are finished,
+and its working tree was clean and synchronized at handoff. Future substrate
+installation remains with railiance-cluster; production approval remains a
+separate evidence gate.
+