Review hall PQRST corpus and close published-record pilot

Assistant: codex
Assistant-Model: gpt-6-astra
Assistant-Session: 01a0e759-301a-78b1-bbc1-040ef094b12d

PqrstPrompt.md

@@ -6,6 +6,10 @@ Specification: [`spec/PqrstEstimationPractice.md`](spec/PqrstEstimationPractice.
 
 Do not add scoring dimensions inside PQRST. Do not run this prompt mid-session unless deliberately closing a phase.
 
+Estimate the substantive work before the closing ritual. Writing the hall entry,
+rendering its portrait, and the final hall sync are excluded. Coordination and
+verification performed as part of the substantive task still count.
+
 ---
 
 ```text
@@ -73,7 +77,7 @@ After the block, also emit the equivalent YAML record with keys p, q, r, s, t, t
 
 ## Accepting or rejecting a result
 
-Reject and re-run if any of the following is true:
+Reject the result as a valid record if any of the following is true:
 
 - the five values are not integers summing to 100;
 - `Confidence` is missing or not one of `low` / `medium` / `high`;
@@ -81,3 +85,9 @@ Reject and re-run if any of the following is true:
 - `Dominant factors` merely restates the numbers or says something like "mixed work across several areas";
 - a sixth dimension has been introduced inside the 5-tuple;
 - S is non-zero but no security-specific work actually occurred.
+
+Preserve the original output and the rejection reason. Do not silently repair
+it or re-run for a better answer; a rejected first attempt is useful evidence.
+If an existing record already has a retry, retain both and label their order.
+Use optional `Notes` to preserve attribution uncertainty, limited session
+context, or any prior allocation hint; do not call those records uncoached.