chore: sync workflow templates

.github/scripts/agent_delegation_policy.js

@@ -246,6 +246,7 @@ function decideNextAgent({
       reason: explicitAgent ? 'explicit-label' : 'default',
       shouldSwitch: false,
       alternatives: [],
+      delegationSource: 'static',
     };
   }
 
@@ -282,6 +283,7 @@ function decideNextAgent({
       reason: 'no-agents-available',
       shouldSwitch: false,
       alternatives: [],
+      delegationSource: 'static',
     };
   }
 
@@ -294,6 +296,7 @@ function decideNextAgent({
       reason: 'initial-selection',
       shouldSwitch: false,
       alternatives: availableAgents.filter((a) => a !== initialAgent),
+      delegationSource: 'static',
     };
   }
 
@@ -306,6 +309,7 @@ function decideNextAgent({
       shouldSwitch: true,
       previousAgent: currentAgent,
       alternatives: availableAgents.filter((agent) => agent !== nextAgent),
+      delegationSource: 'static',
     };
   }
 
@@ -325,6 +329,7 @@ function decideNextAgent({
       reason: `effective (${effectiveness.summary})`,
       shouldSwitch: false,
       alternatives: availableAgents.filter((a) => a !== currentAgent),
+      delegationSource: 'static',
     };
   }
 
@@ -335,6 +340,7 @@ function decideNextAgent({
       reason: `cooldown (${5 - roundsSinceSwitch} rounds remaining)`,
       shouldSwitch: false,
       alternatives: availableAgents.filter((a) => a !== currentAgent),
+      delegationSource: 'static',
     };
   }
 
@@ -373,7 +379,7 @@ function decideNextAgent({
         reason: `stalled-no-alternatives (${sourceSuffix})`,
         shouldSwitch: false,
         alternatives: [],
-        delegationSource: 'static',
+        delegationSource,
       };
     }
     core?.info?.(`Switching from ${currentAgent} to ${nextAgent} due to stall (${sourceSuffix})`);

scripts/langchain/issue_dedup.py

@@ -6,6 +6,7 @@
 from __future__ import annotations
 
 import logging
+import math
 import os
 from collections.abc import Iterable, Mapping
 from dataclasses import dataclass
@@ -198,16 +199,19 @@ def find_similar_issues(
     min_score = _resolve_threshold(threshold)
     matches: list[IssueMatch] = []
     for doc, raw_score in results:
+        float_score = float(raw_score)
+        if not math.isfinite(float_score):
+            continue
         metadata = getattr(doc, "metadata", {}) or {}
         fallback_title = getattr(doc, "page_content", None)
         issue = _issue_from_metadata(metadata, fallback_title)
-        similarity = _similarity_from_score(float(raw_score), score_type)
+        similarity = _similarity_from_score(float_score, score_type)
         if similarity >= min_score:
             matches.append(
                 IssueMatch(
                     issue=issue,
                     score=similarity,
-                    raw_score=float(raw_score),
+                    raw_score=float_score,
                     score_type=score_type,
                 )
             )
@@ -217,6 +221,9 @@ def find_similar_issues(
 
 
 def _format_similarity(score: float) -> str:
+    """Format finite scores as percentages; invalid scores provide no similarity."""
+    if not math.isfinite(score):
+        return "0%"
     clamped = min(max(score, 0.0), 1.0)
     return f"{round(clamped * 100):d}%"