diff --git a/main.py b/main.py
index 63d6e4e..a6dc493 100644
--- a/main.py
+++ b/main.py
@@ -19,7 +19,8 @@ from schemas import (
     Convention as ConventionSchema,
     MemoryBase,
     Memory as MemorySchema,
-    ContextBundle
+    ContextBundle,
+    CompressionRequest
 )
 from rag import build_context_bundle, clear_cache as clear_rag_cache
 from compression import compress_conversation, count_tokens
@@ -442,7 +443,7 @@ async def compress_messages_endpoint(
         ollama_model=config.compression.ollama_model,
         ollama_url=config.compression.ollama_url
     )
-    original_tokens = sum(count_tokens(m.get("content", "")) for m in messages)
+    original_tokens = sum(count_tokens(m.get("content", "")) for m in request.messages)
     compressed_tokens = sum(count_tokens(m.get("content", "")) for m in compressed)
     
     return {