fix(retrieval): the completion-report arm gets its own bar, not the prompt arm's (#3860)

`report_preference` shipped reading PROMPTRULE_THRESHOLD_KEY, so the two arms
were one dial: tuning the bar for an operator's prose silently retuned the
lookup that runs when a task closes.

That coupling is worse on this arm than it would be anywhere else. Every other
retrieval arm scores a query that varies per call, so a mis-set bar shows up as
a changed clear-rate. COMPLETION_QUERY is a fixed string, so this arm's best
score for a given corpus is a CONSTANT — and a constant sitting under the bar is
a dead arm rather than a quiet one. No volume of traffic reveals it.

Found by the first live read for milestone 394 step 9: 69 calls, 69 declines,
every one naming the same record at the same score (0.7194 against a 0.72 bar).
Reading `best_available_id` (#3807) showed the record was about interpreting a
REQUEST, not about report shape — so the declines were correct and the arm is
healthy. The percentile alone would have said "lower the bar", which would have
delivered a false positive on every completion report ever written.

The bar does not move; the key does. Both defaults stay 0.72, so this changes no
behaviour on any install — it makes "leave this one where it is" expressible,
which it was not before. Settings grows the control (rules 25, 27), and the
default-agreement check grows a row.

Deliberately NOT included: a change to PROMPTRULE_DEFAULT_THRESHOLD. The
evidence for moving it is this install's near-miss table, and rule 115 keeps a
shipped default from being justified by one instance's corpus. That bar is a
per-user setting and belongs in the operator's Settings, not in the product.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01821k5B3Ysecp9fNYs92Kuy

frontend/src/views/SettingsView.vue

@@ -96,6 +96,7 @@ const kbToolRuleThreshold = ref("0.68");
 // And the prompt boundary is a third query shape again — the operator's own
 // prose rather than anything a tool produced (#3852).
 const kbPromptRuleThreshold = ref("0.72");
+const kbReportPrefThreshold = ref("0.72");
 // Near-duplicate report floors, one per record kind (services/dedup.py).
 // Snippets are single-chunk, so their floor sits below the 0.90 write-time
 // gate and catches what it lets through. Notes/tasks are scored at chunk
@@ -171,6 +172,7 @@ async function saveKbInject() {
   // Bash call, so a fallback of 0 would put a rule in front of every command.
   const trT = Math.min(1, Math.max(0, Number(kbToolRuleThreshold.value) || 0.68));
   const prT = Math.min(1, Math.max(0, Number(kbPromptRuleThreshold.value) || 0.72));
+  const rpT = Math.min(1, Math.max(0, Number(kbReportPrefThreshold.value) || 0.72));
   kbInjectThreshold.value = String(t);
   kbInjectTopK.value = String(k);
   kbDupThresholdSnippet.value = String(dupSnip);
@@ -181,6 +183,7 @@ async function saveKbInject() {
   kbRuleHintThreshold.value = String(rhT);
   kbToolRuleThreshold.value = String(trT);
   kbPromptRuleThreshold.value = String(prT);
+  kbReportPrefThreshold.value = String(rpT);
   savingKbInject.value = true;
   kbInjectSaved.value = false;
   try {
@@ -202,6 +205,12 @@ async function saveKbInject() {
       // queries are different shapes. Moving one must not move the others.
       kb_toolrule_threshold: String(trT),
       kb_promptrule_threshold: String(prT),
+      // A SIXTH, and the one that most needed its own key: this arm's query
+      // is a fixed string, so its score is a constant for a given corpus.
+      // While it borrowed the prompt bar, tuning prose silently retuned it —
+      // and a constant that lands under the bar is a dead arm, not a quiet
+      // one (#3860).
+      kb_reportpref_threshold: String(rpT),
       kb_duplicate_threshold_snippet: String(dupSnip),
       kb_duplicate_threshold_note: String(dupNote),
       kb_duplicate_threshold_task: String(dupTask),
@@ -657,6 +666,9 @@ onMounted(async () => {
   if (allSettings.kb_promptrule_threshold !== undefined) {
     kbPromptRuleThreshold.value = allSettings.kb_promptrule_threshold;
   }
+  if (allSettings.kb_reportpref_threshold !== undefined) {
+    kbReportPrefThreshold.value = allSettings.kb_reportpref_threshold;
+  }
   if (allSettings.kb_writepath_threshold !== undefined) {
     kbWritePathThreshold.value = allSettings.kb_writepath_threshold;
   }
@@ -1568,6 +1580,28 @@ async function deleteUser(userId: number) {
             not a command or a file — which is why it carries its own number.
           </p>
         </div>
+        <div class="field">
+          <label for="kb-reportpref-threshold">Completion-report confidence threshold (0–1)</label>
+          <input
+            id="kb-reportpref-threshold"
+            v-model="kbReportPrefThreshold"
+            type="number"
+            min="0"
+            max="1"
+            step="0.01"
+            class="fs-input input"
+            style="max-width: 8rem"
+          />
+          <p class="field-hint">
+            The bar for a <em>preference about how a completion report should be
+            written</em>, looked up when a task closes. Unlike every other bar
+            here, the question this arm asks never changes — so its score is
+            fixed by your preferences alone, and it will either always find one
+            or never find one. If you have written a preference for report shape
+            and it is not arriving, lower this; there is no run of calls that
+            will reveal the problem on its own.
+          </p>
+        </div>
         <!-- A design system belongs to a PROJECT, and the picker for it lives on
              the project. There was a setting here that designated the system
              this install's own interface was built from; it only ever described

src/scribe/services/plugin_context.py

@@ -471,6 +471,25 @@ PROMPTRULE_DEFAULT_THRESHOLD = 0.72
 # at all, where the act arms never face it because a command is one thing.
 PROMPTRULE_LIMIT = 3
 
+# THE COMPLETION-REPORT ARM'S OWN BAR (services/reply_preferences.py).
+#
+# It borrowed PROMPTRULE_THRESHOLD_KEY when it shipped, which made the two
+# arms one dial: an operator lowering the bar for their own prose moved this
+# one with it, silently. That contradicts the rule every other bar here
+# follows — one number cannot serve arms whose queries are different shapes —
+# and this arm's query is the most different of all. The others score an
+# operator's prose or a session's code, both of which vary per call; this one
+# scores a FIXED string (`COMPLETION_QUERY`) against rule triggers, so its
+# score for a given corpus is a constant. A constant that lands under the bar
+# is not a quiet arm, it is a dead one, and nothing about the prose arm's
+# traffic would ever reveal it.
+#
+# Kept at the prose arm's starting value rather than tuned: the split is what
+# makes the two independently movable, and a default is a product decision
+# that this install's corpus cannot settle (rule 115).
+REPORTPREF_THRESHOLD_KEY = "kb_reportpref_threshold"
+REPORTPREF_DEFAULT_THRESHOLD = 0.72
+
 
 def _slugify(text: str) -> str:
     """kebab-case slug for a skill directory name (a-z0-9 + single hyphens)."""

src/scribe/services/reply_preferences.py

@@ -34,14 +34,26 @@ delivery entirely in retrieval, as 394 decided, and leaves an operator nothing
 new to learn: a preference reaches the completion report the same way every
 other record reaches its moment.
 
-THE BAR IS THE PROMPT ARM'S, AND IT IS NOT YET EARNED HERE
+THE BAR IS ITS OWN, AND THE FIRST READING EARNED IT (#3860)
 
-A fixed query against triggers is a different score distribution from an
-operator's message against the same documents. Starting at the prompt arm's
-setting is the value with evidence behind it, and an operator's tuning of that
-bar reaches this too. Every call logs under its own source,
-`report_preference`, so step 6 can read this surface's near misses apart from
-the prompt arm's before anyone moves the number.
+This borrowed the prompt arm's key when it shipped, on the argument that a
+fixed query against triggers is a different score distribution from an
+operator's message against the same documents — true, and the reason the two
+could not stay one dial. Five days of traffic settled it.
+
+What the readout said: 69 calls, 69 declines, every one naming the SAME record
+at the SAME score (rule 77 at 0.7194 against a 0.72 bar). That constancy is
+the signature of this arm — `COMPLETION_QUERY` never varies, so for a given
+corpus its best score is a constant, and a constant sitting under the bar is a
+dead arm rather than a quiet one. The record it kept declining was about
+reading a REQUEST, not about the shape of a report, so the decline was right
+and the arm is healthy: this install simply has no completion-report
+preference on file.
+
+The bar stayed at 0.72, and the key moved out (REPORTPREF_THRESHOLD_KEY) so
+that staying is a decision rather than a side effect of what the prose arm is
+set to. A surface whose score cannot vary is the one surface where a borrowed
+bar can be wrong forever without a single call looking unusual.
 """
 from __future__ import annotations
 
@@ -50,8 +62,8 @@ import time
 
 from scribe.services.embeddings import semantic_search_rules
 from scribe.services.plugin_context import (
-    PROMPTRULE_DEFAULT_THRESHOLD,
-    PROMPTRULE_THRESHOLD_KEY,
+    REPORTPREF_DEFAULT_THRESHOLD,
+    REPORTPREF_THRESHOLD_KEY,
 )
 from scribe.services.retrieval_telemetry import record_retrieval
 from scribe.services.rule_usage import record_rule_surfaced
@@ -79,9 +91,9 @@ LIMIT = 3
 async def _threshold(user_id: int) -> float:
     try:
         value = float(await get_setting(
-            user_id, PROMPTRULE_THRESHOLD_KEY, str(PROMPTRULE_DEFAULT_THRESHOLD)))
+            user_id, REPORTPREF_THRESHOLD_KEY, str(REPORTPREF_DEFAULT_THRESHOLD)))
     except (TypeError, ValueError):
-        value = PROMPTRULE_DEFAULT_THRESHOLD
+        value = REPORTPREF_DEFAULT_THRESHOLD
     return min(1.0, max(0.0, value))
 
 

tests/test_services_reply_preferences.py

@@ -73,6 +73,46 @@ def test_it_is_a_ranked_source():
     assert not is_ambient("report_preference")
 
 
+def test_its_bar_is_its_own_key_not_the_prompt_arm_s(monkeypatch):
+    """Regression on the coupling #3860 found (and on the fix being real).
+
+    This arm shipped reading PROMPTRULE_THRESHOLD_KEY, so an operator tuning
+    the bar for their own prose moved this one with it and was never told.
+    That is worse here than anywhere else: every other arm scores a query that
+    varies per call, while COMPLETION_QUERY is fixed — so this arm's score for
+    a given corpus is a CONSTANT, and a constant that lands under the bar is a
+    dead arm rather than a quiet one. No amount of traffic reveals it.
+
+    Asserted on the key the lookup actually asks for, which is the thing that
+    broke, rather than on the constant being defined somewhere.
+    """
+    import asyncio
+
+    from scribe.services import plugin_context
+    from scribe.services.reply_preferences import completion_preferences
+
+    asked: list[str] = []
+
+    async def get_setting(user_id, key, default):
+        asked.append(key)
+        return default
+
+    async def search(user_id, query, **kw):
+        kw["report"].update({"searched": True})
+        return []
+
+    with patch(f"{M}.get_setting", AsyncMock(side_effect=get_setting)), \
+         patch(f"{M}.semantic_search_rules", AsyncMock(side_effect=search)), \
+         patch(f"{M}.record_retrieval"):
+        asyncio.run(completion_preferences(7))
+
+    assert asked == [plugin_context.REPORTPREF_THRESHOLD_KEY]
+    assert plugin_context.REPORTPREF_THRESHOLD_KEY != plugin_context.PROMPTRULE_THRESHOLD_KEY, (
+        "the two keys are the same string again, so the settings form has one "
+        "dial driving two arms — which is the defect, whatever the value is"
+    )
+
+
 def test_the_query_assumes_no_particular_domain():
     from scribe.services.reply_preferences import COMPLETION_QUERY
 

tests/test_settings_defaults_agree.py

@@ -50,6 +50,7 @@ _PAIRS = (
     ("plugin_context.py", "RULEHINT_DEFAULT_THRESHOLD", "kbRuleHintThreshold"),
     ("plugin_context.py", "TOOLRULE_DEFAULT_THRESHOLD", "kbToolRuleThreshold"),
     ("plugin_context.py", "PROMPTRULE_DEFAULT_THRESHOLD", "kbPromptRuleThreshold"),
+    ("plugin_context.py", "REPORTPREF_DEFAULT_THRESHOLD", "kbReportPrefThreshold"),
     # The plan gate (milestone 415): it blocks a create, so a form showing a
     # looser bar than the one in force would be the more misleading drift.
     ("dedup.py", "PLAN_MATCH_DEFAULT_THRESHOLD", "kbPlanMatchThreshold"),