skip content < 100 chars

mcp/server.py

@@ -35,18 +35,12 @@ mcp = FastMCP("Cavepedia MCP")
 def get_user_roles() -> list[str]:
     """Extract user roles from the X-User-Roles header."""
     headers = get_http_headers()
-    print(f"DEBUG: All headers received: {dict(headers)}")
     roles_header = headers.get("x-user-roles", "")
-    print(f"DEBUG: x-user-roles header value: '{roles_header}'")
     if roles_header:
         try:
-            roles = json.loads(roles_header)
+            return json.loads(roles_header)
-            print(f"DEBUG: Parsed roles: {roles}")
+        except json.JSONDecodeError:
-            return roles
-        except json.JSONDecodeError as e:
-            print(f"DEBUG: JSON decode error: {e}")
             return []
-    print("DEBUG: No roles header found, returning empty list")
     return []
 
 def is_sources_only() -> bool:
@@ -82,7 +76,7 @@ def search_caving_documents(query: str, priority_prefixes: list[str] | None = No
     top_n = 2
     candidate_limit = top_n * 4
     rows = conn.execute(
-        'SELECT * FROM embeddings WHERE embedding IS NOT NULL AND role = ANY(%s) ORDER BY embedding <=> %s::vector LIMIT %s',
+        'SELECT * FROM embeddings WHERE embedding IS NOT NULL AND LENGTH(content) > 100 AND role = ANY(%s) ORDER BY embedding <=> %s::vector LIMIT %s',
         (roles, query_embedding, candidate_limit)
     ).fetchall()
 
@@ -102,6 +96,7 @@ def search_caving_documents(query: str, priority_prefixes: list[str] | None = No
     sources_only = is_sources_only()
     for result in rerank_resp.results:
         row = rows[result.index]
+        content = row['content'] or ''
         score = result.relevance_score
 
         # Boost score if key starts with any priority prefix (e.g., 'nss/aca')
@@ -113,7 +108,6 @@ def search_caving_documents(query: str, priority_prefixes: list[str] | None = No
         if sources_only:
             docs.append({'key': row['key'], 'relevance': round(score, 3)})
         else:
-            content = row['content'] or ''
             docs.append({'key': row['key'], 'content': content, 'relevance': round(score, 3)})
 
     # Re-sort by boosted score and return top_n