diff --git a/main.py b/main.py
index 0f12e7a..63d6e4e 100644
--- a/main.py
+++ b/main.py
@@ -422,7 +422,7 @@ async def get_context_rag(
 
 @app.post("/compress", dependencies=[Depends(verify_api_key)])
 async def compress_messages_endpoint(
-    messages: List[dict],
+    request: CompressionRequest,
     keep_last_n: Optional[int] = None,
     max_tokens: Optional[int] = None
 ):
@@ -435,9 +435,9 @@ async def compress_messages_endpoint(
     max_tokens = max_tokens or config.compression.max_tokens
     
     compressed = await compress_conversation(
-        messages,
-        max_tokens=max_tokens,
-        keep_last_n=keep_last_n,
+        request.messages,
+        max_tokens=max_tokens or config.compression.max_tokens,
+        keep_last_n=keep_last_n or config.compression.keep_last_n,
         strategy=config.compression.strategy,
         ollama_model=config.compression.ollama_model,
         ollama_url=config.compression.ollama_url
diff --git a/schemas.py b/schemas.py
index 418b2b0..f51d5ed 100644
--- a/schemas.py
+++ b/schemas.py
@@ -67,6 +67,17 @@ class Memory(MemoryBase):
         from_attributes = True
 
 
+class Message(BaseModel):
+    role: str
+    content: str
+
+
+class CompressionRequest(BaseModel):
+    messages: List[Message]
+    keep_last_n: Optional[int] = None
+    max_tokens: Optional[int] = None
+
+
 class ContextBundle(BaseModel):
     skills: List[Skill]
     snippets: List[Snippet]