fix(resolvers): question-log session_id becomes a substitution placeholder + stale-comment sweep

The question-log block bound $_SESSION_ID, a shell variable the
consolidated fence never sets — hook-less hosts logged empty session_id,
breaking /plan-tune per-session grouping. It now uses the same
substitute-from-the-skill-start-echoes contract as the telemetry block.
Also: retired the pre-Phase-2 stop-gate docstring, repointed the
gbrain-local-status cross-reference at the script's inline jq, dropped an
orphaned section comment, documented retro-metrics' suffix-only census.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

bin/gstack-retro-metrics

@@ -297,6 +297,9 @@ echo "PRS_REFERENCED: $(printf '%s' "$_PRS" | wc -w | tr -d ' ')"
 [ -n "$_PRS" ] && echo "PR_REFS: $_PRS" || true
 
 # ── Test health (repo-wide + window) ─────────────────────────────────────────
+# Deliberately suffix-only (narrower than is_test's dir-based patterns): the
+# repo-wide census counts conventional test FILES; is_test additionally counts
+# dir-homed helpers toward test-insertion ratios.
 _TF_TOTAL=$(git ls-files 2>/dev/null | grep -cE '(\.test\.|\.spec\.|_test\.|_spec\.)' || true)
 case "$_TF_TOTAL" in ''|*[!0-9]*) _TF_TOTAL=0 ;; esac
 echo "TEST_FILES_TOTAL: $_TF_TOTAL"

lib/gbrain-local-status.ts

@@ -146,9 +146,9 @@ function gbrainConfigPath(env?: NodeJS.ProcessEnv): string {
  * cwd's NEAREST-ANCESTOR project scope only (#2499 made project scope
  * visible; the per-project scoping fixes the machine-wide bleed where ONE
  * project's remote registration reclassified broken local engines as
- * thin-client for EVERY cwd). Ancestor matching mirrors the
- * GBRAIN_MCP_ENTRY_JQ resolution in
- * scripts/resolvers/preamble/generate-brain-sync-block.ts: cwd == key or
+ * thin-client for EVERY cwd). Ancestor matching mirrors the inline jq
+ * resolution in bin/gstack-skill-start (the single bash copy of the #2499
+ * resolution, moved there in token-reduction Phase 1): cwd == key or
  * cwd startswith key + separator, longest matching key that actually
  * carries a gbrain entry wins (a nested project WITHOUT gbrain doesn't
  * shadow its parent's registration).

scripts/resolvers/preamble.ts

@@ -28,8 +28,6 @@ import {
   generatePlanModeInfo,
 } from './preamble/generate-completion-status';
 
-// One-time onboarding prompts
-
 // Host-specific instructions
 import { generateBrainHealthInstruction } from './preamble/generate-brain-health-instruction';
 

scripts/resolvers/preamble/generate-brain-sync-block.ts

@@ -6,9 +6,10 @@
  * daily pull, queue-depth status) now runs inside `bin/gstack-skill-start`
  * (invoked by the Preamble fence above) and reports through the same STATUS
  * lines it always emitted: the GBrain hint text and the `ARTIFACTS_SYNC:`
- * line. What remains here is the prose the model acts on — including the
- * one-time privacy stop-gate, which stays inline until the Phase 2
- * instruction-emission layer moves it behind its runtime gate.
+ * line. What remains here is the prose the model acts on. The one-time
+ * privacy stop-gate moved behind its runtime gate in Phase 2: it now arrives
+ * as a GSTACK_INSTRUCTION block from gstack-skill-start when consent is
+ * actually pending; this generator only carries the interpretation prose.
  *
  * Skill-END sync is no longer a separate fence: `bin/gstack-skill-end`
  * (invoked by the Telemetry step) drains the queue before logging.

scripts/resolvers/question-tuning.ts

@@ -33,9 +33,9 @@ Before each AskUserQuestion, choose \`question_id\` from \`${registry}\` or \`{s
 
 **Embed the option recommendation via the \`(recommended)\` label suffix** on exactly one option per AUQ. The PreToolUse hook parses \`(recommended)\` first, falls back to "Recommendation: X" prose, and refuses to auto-decide if ambiguous. Two \`(recommended)\` labels = refuse.
 
-After answer, log best-effort (PostToolUse hook also captures deterministically when installed; dedup on (source, tool_use_id) handles double-writes):
+After answer, log best-effort (PostToolUse hook also captures deterministically when installed; dedup on (source, tool_use_id) handles double-writes). Substitute \`SESSION_ID\` with the value the preamble's skill-start output echoed — shell variables do not survive between Bash calls:
 \`\`\`bash
-${bin}/gstack-question-log '{"skill":"${ctx.skillName}","question_id":"<id>","question_summary":"<short>","category":"<approval|clarification|routing|cherry-pick|feedback-loop>","door_type":"<one-way|two-way>","options_count":N,"user_choice":"<key>","recommended":"<key>","session_id":"'"$_SESSION_ID"'"}' 2>/dev/null || true
+${bin}/gstack-question-log '{"skill":"${ctx.skillName}","question_id":"<id>","question_summary":"<short>","category":"<approval|clarification|routing|cherry-pick|feedback-loop>","door_type":"<one-way|two-way>","options_count":N,"user_choice":"<key>","recommended":"<key>","session_id":"SESSION_ID"}' 2>/dev/null || true
 \`\`\`
 
 For two-way questions, offer: "Tune this question? Reply \`tune: never-ask\`, \`tune: always-ask\`, or free-form."