From 83ca87549252a3bb28988c9351e0a6e79edcd64f Mon Sep 17 00:00:00 2001
From: cybersecurity-cell <kalai.kandasamy@gmail.com>
Date: Wed, 12 Aug 2026 05:43:01 -0400
Subject: [PATCH] fix: suggest_questions bridge-node question restated hub's
 own name as a target

An extreme hub node is often isolated by Louvain into its own singleton
community, auto-named after that same node. The bridge-node question
template named that community right after the node itself, producing
"Why does `useLang()` connect `useLang` to `Settings`, `Chart`?" - the
node's own name shows up twice, once as a distinct "target" it supposedly
connects to.

Drop the redundant "connect `{comm_label}` to" framing when the bridge
node's own community label normalizes to the same string as the node's
own label, and use "Why does `{label}` bridge {other_labels}?" instead.
Leaves the phrasing unchanged when the two labels genuinely differ.

Found while tracing a suggested question on a real 3000+ node graph,
where 3 of the top-3 bridge questions all hit this (the highest-degree
nodes are exactly the ones most likely to end up in their own
singleton community).
---
 graphify/analyze.py   | 13 ++++++++++++-
 tests/test_analyze.py | 23 +++++++++++++++++++++++
 2 files changed, 35 insertions(+), 1 deletion(-)

diff --git a/graphify/analyze.py b/graphify/analyze.py
index 0707e2be7..0365eb5e2 100644
--- a/graphify/analyze.py
+++ b/graphify/analyze.py
@@ -1,5 +1,6 @@
 """Graph analysis: god nodes (most connected), surprising connections (cross-community), suggested questions."""
 from __future__ import annotations
+import re
 from pathlib import Path
 import networkx as nx
 
@@ -473,9 +474,19 @@ def suggest_questions(
             neighbor_comms = {node_community.get(n) for n in neighbors if node_community.get(n) != cid}
             if neighbor_comms:
                 other_labels = [community_labels.get(c, f"Community {c}") for c in neighbor_comms]
+                others = ', '.join(f'`{l}`' for l in other_labels)
+                # A bridge node's own community is sometimes a singleton auto-named
+                # after that same node (an extreme hub gets isolated into its own
+                # community by Louvain), which read as "Why does `X` connect `X`
+                # to ...?" - restating the node's own name as a distinct target.
+                _norm = lambda s: re.sub(r"[^a-z0-9]", "", str(s).lower())
+                if _norm(comm_label) == _norm(label):
+                    question_text = f"Why does `{label}` bridge {others}?"
+                else:
+                    question_text = f"Why does `{label}` connect `{comm_label}` to {others}?"
                 questions.append({
                     "type": "bridge_node",
-                    "question": f"Why does `{label}` connect `{comm_label}` to {', '.join(f'`{l}`' for l in other_labels)}?",
+                    "question": question_text,
                     "why": f"High betweenness centrality ({score:.3f}) - this node is a cross-community bridge.",
                 })
 
diff --git a/tests/test_analyze.py b/tests/test_analyze.py
index 7bff432cf..2f93680ce 100644
--- a/tests/test_analyze.py
+++ b/tests/test_analyze.py
@@ -616,6 +616,29 @@ def test_suggest_questions_excludes_rationale_nodes_from_isolated_count():
     assert "Explains service" not in isolated["question"]
 
 
+def test_suggest_questions_bridge_node_no_self_referential_connect():
+    """An extreme hub often gets isolated into its own singleton community,
+    auto-named after that same node. The bridge-node question must not restate
+    that as "Why does `X` connect `X` to ...?" - `X` is not a distinct target."""
+    G = nx.Graph()
+    G.add_node("hub", label="useLang", source_file="lib/i18n/context.tsx", file_type="code")
+    G.add_node("peer_a", label="SettingsContent", source_file="app/settings/settings-content.tsx", file_type="code")
+    G.add_node("peer_b", label="ChartContent", source_file="app/chart/chart-content.tsx", file_type="code")
+    G.add_edge("hub", "peer_a")
+    G.add_edge("hub", "peer_b")
+
+    communities = {0: ["hub"], 1: ["peer_a"], 2: ["peer_b"]}
+    community_labels = {0: "useLang", 1: "Settings", 2: "Chart"}
+
+    questions = suggest_questions(G, communities, community_labels, top_n=10)
+    bridge = next(question for question in questions if question["type"] == "bridge_node")
+
+    assert "connect `useLang`" not in bridge["question"]
+    assert "bridge" in bridge["question"]
+    assert "`Settings`" in bridge["question"]
+    assert "`Chart`" in bridge["question"]
+
+
 # ── find_import_cycles tests ──────────────────────────────────────────────────