@miller-tech/uap 1.20.40 → 1.20.41

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@miller-tech/uap",
-  "version": "1.20.40",
+  "version": "1.20.41",
   "description": "Autonomous AI agent memory system with CLAUDE.md protocol enforcement",
   "type": "module",
   "main": "dist/index.js",

package/tools/agents/scripts/anthropic_proxy.py

@@ -8250,7 +8250,7 @@ async def models():
             {"id": "claude-haiku-4-5-20251001", "object": "model"},
             {"id": "claude-sonnet-4-6", "object": "model"},
             {"id": "claude-opus-4-7", "object": "model"},
-            {"id": "qwen35-a3b-iq4xs", "object": "model"},
+            {"id": "qwen36-27b-iq4xs", "object": "model"},
         ]
     }
 

package/tools/agents/tests/test_anthropic_proxy_streaming.py

@@ -5028,9 +5028,13 @@ class TestModelsEndpoint(unittest.TestCase):
         self.assertIn("claude-sonnet-4-6", ids)
         self.assertIn("claude-opus-4-7", ids)
 
-        # Local model (kept because requests for it actually route locally
-        # even with __local_only__ passthrough sentinel set)
-        self.assertIn("qwen35-a3b-iq4xs", ids)
+        # Local model — what llama-server actually serves. Updated
+        # 2026-05-15 from qwen35-a3b-iq4xs after the switch from 35B-A3B
+        # MoE to Qwen3.6-27B dense (see project_active_server memory).
+        # Requests for this ID route locally even with __local_only__
+        # passthrough sentinel set.
+        self.assertIn("qwen36-27b-iq4xs", ids)
+        self.assertNotIn("qwen35-a3b-iq4xs", ids)
 
     def test_models_endpoint_drops_stale_4_6_dated_variants(self):
         """The pre-2026-05 list advertised claude-opus-4-6-20260101,