From 795462757772ea5ac8bceee130beae8d69ee77d0 Mon Sep 17 00:00:00 2001
From: Cameron Pfiffer <cameron@pfiffer.org>
Date: Wed, 28 Jan 2026 15:38:34 -0800
Subject: [PATCH] Fix bare URL pattern to not match parts of failed mentions

---
 data/published_messages.txt | 5 +++++
 tools/agent.py              | 3 ++-
 2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/data/published_messages.txt b/data/published_messages.txt
index 8d73eec..926db24 100644
--- a/data/published_messages.txt
+++ b/data/published_messages.txt
@@ -850,3 +850,8 @@ message-aa847d16-06e6-4494-b6ae-97d662de4ad6
 message-3f6149bc-a7e2-4e5b-868a-eb69de18529e
 message-f5e6b3a0-1177-4655-9aae-6568a0e4b29a
 message-3608bf83-c59f-449f-9dda-f1277044a792
+message-259334cb-1e9d-4a53-942b-3971a36c9f10
+message-d6331c1f-a0ec-4b79-b664-32aed7587b3d
+message-d6331c1f-a0ec-4b79-b664-32aed7587b3d
+message-d21c51c4-41f7-437a-b14c-956f39fb0d7b
+message-386d1893-a2a0-4bca-bf90-5bf748c17e7c
diff --git a/tools/agent.py b/tools/agent.py
index c412558..001d6a5 100644
--- a/tools/agent.py
+++ b/tools/agent.py
@@ -141,7 +141,8 @@ async def parse_facets(text: str) -> list:
         })
     
     # Find URLs without protocol (common TLDs)
-    bare_url_pattern = r'(?<![/@\w])([a-zA-Z0-9][-a-zA-Z0-9]*\.(?:com|org|net|io|app|social|network|blue|xyz|dev|ai|bsky\.social)[^\s<>\[\]()\'\"]*)'
+    # Negative lookbehind excludes: after @, after /, after word char, after .
+    bare_url_pattern = r'(?<![/@\w.])([a-zA-Z0-9][-a-zA-Z0-9]*\.(?:com|org|net|io|app|social|network|blue|xyz|dev|ai)[^\s<>\[\]()\'\"]*)'
     for match in re.finditer(bare_url_pattern, text):
         bare_url = match.group(1).rstrip(".,;:!?")
         if not bare_url:
-- 
2.51.2