Feat v0.5.3 chat polish + integration testing

Conversation search (db.query search_like, chat-core /search endpoint,
sidebar search UI), message pagination polish (scroll preservation,
loading spinner), workflow-chat integration package, and multi-user
E2E test infrastructure.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

packages/chat-core/script.star

@@ -1,4 +1,4 @@
-# Chat Core — Starlark Backend (v0.1.0)
+# Chat Core — Starlark Backend (v0.2.0)
 #
 # Library package providing conversations, messages, participants,
 # and read cursors. Consumable via lib.require("chat-core").
@@ -309,6 +309,10 @@ def on_request(req):
     if method == "POST" and path == "/conversations":
         return _handle_create_conversation(req, user_id)
 
+    # GET /search?q=term — search conversations and messages
+    if method == "GET" and path == "/search":
+        return _handle_search(req, user_id)
+
     # GET /unread — unread counts
     if method == "GET" and path == "/unread":
         return _handle_unread(user_id)
@@ -652,6 +656,59 @@ def _handle_mark_read(cid, req, user_id):
     return _resp(200, {"ok": True})
 
 
+def _handle_search(req, user_id):
+    """Search across conversation titles and message content."""
+    q = _str(req.get("query", {}).get("q", ""))
+    if len(q) < 2:
+        return _resp(400, {"error": "query must be at least 2 characters"})
+
+    pattern = "%" + q + "%"
+
+    # Get user's conversation IDs
+    my_parts = db.query("participants", filters={"participant_id": user_id}, limit=500)
+    if not my_parts:
+        return _resp(200, {"conversations": [], "messages": []})
+
+    cid_set = {}
+    for p in my_parts:
+        cid_set[_str(p.get("conversation_id", ""))] = True
+
+    # Search conversations by title
+    matching_convs = db.query("conversations", search_like={"title": pattern}, limit=50)
+    conv_results = []
+    for conv in (matching_convs or []):
+        cid = _str(conv.get("id", ""))
+        if cid in cid_set:
+            conv_results.append({
+                "id": cid,
+                "title": conv.get("title", ""),
+                "type": conv.get("type", ""),
+                "updated_at": conv.get("updated_at", ""),
+                "created_at": conv.get("created_at", ""),
+            })
+
+    # Search messages in user's conversations
+    msg_results = []
+    for cid in cid_set:
+        if not cid:
+            continue
+        msgs = db.query("messages", filters={"conversation_id": cid}, search_like={"content": pattern}, order="-created_at", limit=5)
+        for m in (msgs or []):
+            msg_results.append({
+                "id": m.get("id", ""),
+                "conversation_id": cid,
+                "participant_id": m.get("participant_id", ""),
+                "content": _str(m.get("content", "")),
+                "content_type": _str(m.get("content_type", "")),
+                "created_at": _str(m.get("created_at", "")),
+            })
+
+    # Sort messages by created_at descending, limit to 50
+    msg_results = sorted(msg_results, key=lambda x: x.get("created_at", ""), reverse=True)[:50]
+
+    return _resp(200, {"conversations": conv_results, "messages": msg_results})
+
+
 def _handle_unread(user_id):
     """Get unread counts for all user's conversations."""
     my_parts = db.query("participants", filters={"participant_id": user_id}, limit=500)