chuk-tool-processor 0.6.11__py3-none-any.whl → 0.6.13__py3-none-any.whl

chuk_tool_processor/mcp/transport/sse_transport.py

@@ -2,17 +2,19 @@
 """
 SSE transport for MCP communication.
 
-FIXED: Removed problematic gateway connectivity test that was causing 401 errors.
-The SSE endpoint works perfectly, so we don't need to test the base URL.
+FIXED: Improved health monitoring to avoid false unhealthy states.
+The SSE endpoint works perfectly, so we need more lenient health checks.
 """
+
 from __future__ import annotations
 
 import asyncio
+import contextlib
 import json
+import logging
 import time
 import uuid
-from typing import Dict, Any, List, Optional
-import logging
+from typing import Any
 
 import httpx
 
@@ -24,47 +26,53 @@ logger = logging.getLogger(__name__)
 class SSETransport(MCPBaseTransport):
     """
     SSE transport implementing the MCP protocol over Server-Sent Events.
-    
-    FIXED: Uses SSE endpoint directly without problematic connectivity tests.
+
+    FIXED: More lenient health monitoring to avoid false unhealthy states.
     """
 
-    def __init__(self, url: str, api_key: Optional[str] = None, 
-                 headers: Optional[Dict[str, str]] = None,
-                 connection_timeout: float = 30.0, 
-                 default_timeout: float = 60.0,
-                 enable_metrics: bool = True):
+    def __init__(
+        self,
+        url: str,
+        api_key: str | None = None,
+        headers: dict[str, str] | None = None,
+        connection_timeout: float = 30.0,
+        default_timeout: float = 60.0,
+        enable_metrics: bool = True,
+    ):
         """
         Initialize SSE transport.
         """
-        self.url = url.rstrip('/')
+        self.url = url.rstrip("/")
         self.api_key = api_key
         self.configured_headers = headers or {}
         self.connection_timeout = connection_timeout
         self.default_timeout = default_timeout
         self.enable_metrics = enable_metrics
-        
+
         logger.debug("SSE Transport initialized with URL: %s", self.url)
-        
+
         # Connection state
         self.session_id = None
         self.message_url = None
-        self.pending_requests: Dict[str, asyncio.Future] = {}
+        self.pending_requests: dict[str, asyncio.Future] = {}
         self._initialized = False
-        
+
         # HTTP clients
         self.stream_client = None
         self.send_client = None
-        
+
         # SSE stream management
         self.sse_task = None
         self.sse_response = None
         self.sse_stream_context = None
-        
-        # Health monitoring
+
+        # FIXED: More lenient health monitoring
         self._last_successful_ping = None
         self._consecutive_failures = 0
-        self._max_consecutive_failures = 3
-        
+        self._max_consecutive_failures = 5  # INCREASED: was 3, now 5
+        self._connection_grace_period = 30.0  # NEW: Grace period after initialization
+        self._initialization_time = None  # NEW: Track when we initialized
+
         # Performance metrics
         self._metrics = {
             "total_calls": 0,
@@ -75,43 +83,43 @@ class SSETransport(MCPBaseTransport):
             "last_ping_time": None,
             "initialization_time": None,
             "session_discoveries": 0,
-            "stream_errors": 0
+            "stream_errors": 0,
         }
 
     def _construct_sse_url(self, base_url: str) -> str:
         """Construct the SSE endpoint URL from the base URL."""
-        base_url = base_url.rstrip('/')
-        
-        if base_url.endswith('/sse'):
+        base_url = base_url.rstrip("/")
+
+        if base_url.endswith("/sse"):
             logger.debug("URL already contains /sse endpoint: %s", base_url)
             return base_url
-        
+
         sse_url = f"{base_url}/sse"
         logger.debug("Constructed SSE URL: %s -> %s", base_url, sse_url)
         return sse_url
 
-    def _get_headers(self) -> Dict[str, str]:
+    def _get_headers(self) -> dict[str, str]:
         """Get headers with authentication and custom headers."""
         headers = {
-            'User-Agent': 'chuk-tool-processor/1.0.0',
-            'Accept': 'text/event-stream',
-            'Cache-Control': 'no-cache',
+            "User-Agent": "chuk-tool-processor/1.0.0",
+            "Accept": "text/event-stream",
+            "Cache-Control": "no-cache",
         }
-        
+
         # Add configured headers first
         if self.configured_headers:
             headers.update(self.configured_headers)
-        
+
         # Add API key as Bearer token if provided
         if self.api_key:
-            headers['Authorization'] = f'Bearer {self.api_key}'
-        
+            headers["Authorization"] = f"Bearer {self.api_key}"
+
         return headers
 
     async def _test_gateway_connectivity(self) -> bool:
         """
         Skip connectivity test - we know the SSE endpoint works.
-        
+
         FIXED: The diagnostic proves SSE endpoint works perfectly.
         No need to test base URL that causes 401 errors.
         """
@@ -119,63 +127,58 @@ class SSETransport(MCPBaseTransport):
         return True
 
     async def initialize(self) -> bool:
-        """Initialize SSE connection."""
+        """Initialize SSE connection with improved health tracking."""
         if self._initialized:
             logger.warning("Transport already initialized")
             return True
-        
+
         start_time = time.time()
-        
+
         try:
             logger.debug("Initializing SSE transport...")
-            
+
             # FIXED: Skip problematic connectivity test
             if not await self._test_gateway_connectivity():
                 logger.error("Gateway connectivity test failed")
                 return False
-            
+
             # Create HTTP clients
             self.stream_client = httpx.AsyncClient(
                 timeout=httpx.Timeout(self.connection_timeout),
                 follow_redirects=True,
-                limits=httpx.Limits(max_connections=10, max_keepalive_connections=5)
+                limits=httpx.Limits(max_connections=10, max_keepalive_connections=5),
             )
             self.send_client = httpx.AsyncClient(
                 timeout=httpx.Timeout(self.default_timeout),
                 follow_redirects=True,
-                limits=httpx.Limits(max_connections=10, max_keepalive_connections=5)
+                limits=httpx.Limits(max_connections=10, max_keepalive_connections=5),
             )
-            
+
             # Connect to SSE stream
             sse_url = self._construct_sse_url(self.url)
             logger.debug("Connecting to SSE endpoint: %s", sse_url)
-            
-            self.sse_stream_context = self.stream_client.stream(
-                'GET', sse_url, headers=self._get_headers()
-            )
+
+            self.sse_stream_context = self.stream_client.stream("GET", sse_url, headers=self._get_headers())
             self.sse_response = await self.sse_stream_context.__aenter__()
-            
+
             if self.sse_response.status_code != 200:
                 logger.error("SSE connection failed with status: %s", self.sse_response.status_code)
                 await self._cleanup()
                 return False
-            
+
             logger.debug("SSE streaming connection established")
-            
+
             # Start SSE processing task
-            self.sse_task = asyncio.create_task(
-                self._process_sse_stream(), 
-                name="sse_stream_processor"
-            )
-            
+            self.sse_task = asyncio.create_task(self._process_sse_stream(), name="sse_stream_processor")
+
             # Wait for session discovery
             logger.debug("Waiting for session discovery...")
             session_timeout = 10.0
             session_start = time.time()
-            
+
             while not self.message_url and (time.time() - session_start) < session_timeout:
                 await asyncio.sleep(0.1)
-                
+
                 # Check if SSE task died
                 if self.sse_task.done():
                     exception = self.sse_task.exception()
@@ -183,51 +186,55 @@ class SSETransport(MCPBaseTransport):
                         logger.error(f"SSE task died during session discovery: {exception}")
                         await self._cleanup()
                         return False
-            
+
             if not self.message_url:
                 logger.error("Failed to discover session endpoint within %.1fs", session_timeout)
                 await self._cleanup()
                 return False
-            
+
             if self.enable_metrics:
                 self._metrics["session_discoveries"] += 1
-            
+
             logger.debug("Session endpoint discovered: %s", self.message_url)
-            
+
             # Perform MCP initialization handshake
             try:
-                init_response = await self._send_request("initialize", {
-                    "protocolVersion": "2024-11-05",
-                    "capabilities": {},
-                    "clientInfo": {
-                        "name": "chuk-tool-processor",
-                        "version": "1.0.0"
-                    }
-                }, timeout=self.default_timeout)
-                
-                if 'error' in init_response:
-                    logger.error("MCP initialize failed: %s", init_response['error'])
+                init_response = await self._send_request(
+                    "initialize",
+                    {
+                        "protocolVersion": "2024-11-05",
+                        "capabilities": {},
+                        "clientInfo": {"name": "chuk-tool-processor", "version": "1.0.0"},
+                    },
+                    timeout=self.default_timeout,
+                )
+
+                if "error" in init_response:
+                    logger.error("MCP initialize failed: %s", init_response["error"])
                     await self._cleanup()
                     return False
-                
+
                 # Send initialized notification
                 await self._send_notification("notifications/initialized")
-                
+
+                # FIXED: Set health tracking state
                 self._initialized = True
+                self._initialization_time = time.time()
                 self._last_successful_ping = time.time()
-                
+                self._consecutive_failures = 0  # Reset failure count
+
                 if self.enable_metrics:
                     init_time = time.time() - start_time
                     self._metrics["initialization_time"] = init_time
-                
+
                 logger.debug("SSE transport initialized successfully in %.3fs", time.time() - start_time)
                 return True
-                
+
             except Exception as e:
                 logger.error("MCP handshake failed: %s", e)
                 await self._cleanup()
                 return False
-                
+
         except Exception as e:
             logger.error("Error initializing SSE transport: %s", e, exc_info=True)
             await self._cleanup()
@@ -237,243 +244,248 @@ class SSETransport(MCPBaseTransport):
         """Process the SSE stream for responses and session discovery."""
         try:
             logger.debug("Starting SSE stream processing...")
-            
+
             current_event = None
-            
+
             async for line in self.sse_response.aiter_lines():
                 line = line.strip()
                 if not line:
                     continue
-                
+
                 # Handle event type declarations
-                if line.startswith('event:'):
-                    current_event = line.split(':', 1)[1].strip()
+                if line.startswith("event:"):
+                    current_event = line.split(":", 1)[1].strip()
                     logger.debug("SSE event type: %s", current_event)
                     continue
-                
+
                 # Handle session endpoint discovery
-                if not self.message_url and line.startswith('data:'):
-                    data_part = line.split(':', 1)[1].strip()
-                    
+                if not self.message_url and line.startswith("data:"):
+                    data_part = line.split(":", 1)[1].strip()
+
                     # NEW FORMAT: event: endpoint + data: https://...
-                    if current_event == "endpoint" and data_part.startswith('http'):
+                    if current_event == "endpoint" and data_part.startswith("http"):
                         self.message_url = data_part
-                        
+
                         # Extract session ID from URL if present
-                        if 'session_id=' in data_part:
-                            self.session_id = data_part.split('session_id=')[1].split('&')[0]
+                        if "session_id=" in data_part:
+                            self.session_id = data_part.split("session_id=")[1].split("&")[0]
                         else:
                             self.session_id = str(uuid.uuid4())
-                        
+
                         logger.debug("Session endpoint discovered via event format: %s", self.message_url)
                         continue
-                    
+
                     # OLD FORMAT: data: /messages/... (backwards compatibility)
-                    elif '/messages/' in data_part:
+                    elif "/messages/" in data_part:
                         endpoint_path = data_part
                         self.message_url = f"{self.url}{endpoint_path}"
-                        
+
                         # Extract session ID if present
-                        if 'session_id=' in endpoint_path:
-                            self.session_id = endpoint_path.split('session_id=')[1].split('&')[0]
+                        if "session_id=" in endpoint_path:
+                            self.session_id = endpoint_path.split("session_id=")[1].split("&")[0]
                         else:
                             self.session_id = str(uuid.uuid4())
-                        
+
                         logger.debug("Session endpoint discovered via old format: %s", self.message_url)
                         continue
-                
+
                 # Handle JSON-RPC responses
-                if line.startswith('data:'):
-                    data_part = line.split(':', 1)[1].strip()
-                    
+                if line.startswith("data:"):
+                    data_part = line.split(":", 1)[1].strip()
+
                     # Skip keepalive pings and empty data
-                    if not data_part or data_part.startswith('ping') or data_part in ('{}', '[]'):
+                    if not data_part or data_part.startswith("ping") or data_part in ("{}", "[]"):
                         continue
-                    
+
                     try:
                         response_data = json.loads(data_part)
-                        
+
                         # Handle JSON-RPC responses with request IDs
-                        if 'jsonrpc' in response_data and 'id' in response_data:
-                            request_id = str(response_data['id'])
-                            
+                        if "jsonrpc" in response_data and "id" in response_data:
+                            request_id = str(response_data["id"])
+
                             # Resolve pending request if found
                             if request_id in self.pending_requests:
                                 future = self.pending_requests.pop(request_id)
                                 if not future.done():
                                     future.set_result(response_data)
                                     logger.debug("Resolved request ID: %s", request_id)
-                    
+
                     except json.JSONDecodeError as e:
                         logger.debug("Non-JSON data in SSE stream (ignoring): %s", e)
-                        
+
         except Exception as e:
             if self.enable_metrics:
                 self._metrics["stream_errors"] += 1
             logger.error("SSE stream processing error: %s", e)
-            self._consecutive_failures += 1
+            # FIXED: Don't increment consecutive failures for stream processing errors
+            # These are often temporary and don't indicate connection health
 
-    async def _send_request(self, method: str, params: Dict[str, Any] = None, 
-                           timeout: Optional[float] = None) -> Dict[str, Any]:
+    async def _send_request(
+        self, method: str, params: dict[str, Any] = None, timeout: float | None = None
+    ) -> dict[str, Any]:
         """Send JSON-RPC request and wait for async response via SSE."""
         if not self.message_url:
             raise RuntimeError("SSE transport not connected - no message URL")
-        
+
         request_id = str(uuid.uuid4())
-        message = {
-            "jsonrpc": "2.0",
-            "id": request_id,
-            "method": method,
-            "params": params or {}
-        }
-        
+        message = {"jsonrpc": "2.0", "id": request_id, "method": method, "params": params or {}}
+
         # Create future for async response
         future = asyncio.Future()
         self.pending_requests[request_id] = future
-        
+
         try:
             # Send HTTP POST request
-            headers = {
-                'Content-Type': 'application/json',
-                **self._get_headers()
-            }
-            
-            response = await self.send_client.post(
-                self.message_url, 
-                headers=headers, 
-                json=message
-            )
-            
+            headers = {"Content-Type": "application/json", **self._get_headers()}
+
+            response = await self.send_client.post(self.message_url, headers=headers, json=message)
+
             if response.status_code == 202:
                 # Async response - wait for result via SSE
                 request_timeout = timeout or self.default_timeout
                 result = await asyncio.wait_for(future, timeout=request_timeout)
-                self._consecutive_failures = 0  # Reset on success
+                # FIXED: Only reset failures on successful tool calls, not all requests
+                if method.startswith("tools/"):
+                    self._consecutive_failures = 0
+                    self._last_successful_ping = time.time()
                 return result
             elif response.status_code == 200:
                 # Immediate response
                 self.pending_requests.pop(request_id, None)
-                self._consecutive_failures = 0  # Reset on success
+                # FIXED: Only reset failures on successful tool calls
+                if method.startswith("tools/"):
+                    self._consecutive_failures = 0
+                    self._last_successful_ping = time.time()
                 return response.json()
             else:
                 self.pending_requests.pop(request_id, None)
-                self._consecutive_failures += 1
+                # FIXED: Only increment failures for tool calls, not initialization
+                if method.startswith("tools/"):
+                    self._consecutive_failures += 1
                 raise RuntimeError(f"HTTP request failed with status: {response.status_code}")
-                
-        except asyncio.TimeoutError:
+
+        except TimeoutError:
             self.pending_requests.pop(request_id, None)
-            self._consecutive_failures += 1
+            # FIXED: Only increment failures for tool calls
+            if method.startswith("tools/"):
+                self._consecutive_failures += 1
             raise
         except Exception:
             self.pending_requests.pop(request_id, None)
-            self._consecutive_failures += 1
+            # FIXED: Only increment failures for tool calls
+            if method.startswith("tools/"):
+                self._consecutive_failures += 1
             raise
 
-    async def _send_notification(self, method: str, params: Dict[str, Any] = None):
+    async def _send_notification(self, method: str, params: dict[str, Any] = None):
         """Send JSON-RPC notification (no response expected)."""
         if not self.message_url:
             raise RuntimeError("SSE transport not connected - no message URL")
-        
-        message = {
-            "jsonrpc": "2.0",
-            "method": method,
-            "params": params or {}
-        }
-        
-        headers = {
-            'Content-Type': 'application/json',
-            **self._get_headers()
-        }
-        
-        response = await self.send_client.post(
-            self.message_url,
-            headers=headers,
-            json=message
-        )
-        
+
+        message = {"jsonrpc": "2.0", "method": method, "params": params or {}}
+
+        headers = {"Content-Type": "application/json", **self._get_headers()}
+
+        response = await self.send_client.post(self.message_url, headers=headers, json=message)
+
         if response.status_code not in (200, 202):
             logger.warning("Notification failed with status: %s", response.status_code)
 
     async def send_ping(self) -> bool:
-        """Send ping to check connection health."""
+        """Send ping to check connection health with improved logic."""
         if not self._initialized:
             return False
-        
+
         start_time = time.time()
         try:
             # Use tools/list as a lightweight ping since not all servers support ping
             response = await self._send_request("tools/list", {}, timeout=10.0)
-            
-            success = 'error' not in response
-            
+
+            success = "error" not in response
+
             if success:
                 self._last_successful_ping = time.time()
-                self._consecutive_failures = 0
-            
+                # FIXED: Don't reset consecutive failures here - let tool calls do that
+
             if self.enable_metrics:
                 ping_time = time.time() - start_time
                 self._metrics["last_ping_time"] = ping_time
                 logger.debug("SSE ping completed in %.3fs: %s", ping_time, success)
-            
+
             return success
         except Exception as e:
             logger.debug("SSE ping failed: %s", e)
-            self._consecutive_failures += 1
+            # FIXED: Don't increment consecutive failures for ping failures
             return False
 
     def is_connected(self) -> bool:
-        """Check if the transport is connected and ready."""
+        """
+        FIXED: More lenient connection health check.
+
+        The diagnostic shows the connection works fine, so we need to be less aggressive
+        about marking it as unhealthy.
+        """
         if not self._initialized or not self.session_id:
             return False
-        
-        # Check if we've had too many consecutive failures
+
+        # FIXED: Grace period after initialization - always return True for a while
+        if self._initialization_time and time.time() - self._initialization_time < self._connection_grace_period:
+            logger.debug("Within grace period - connection considered healthy")
+            return True
+
+        # FIXED: More lenient failure threshold
         if self._consecutive_failures >= self._max_consecutive_failures:
-            logger.warning(f"Connection marked unhealthy after {self._consecutive_failures} failures")
+            logger.warning(f"Connection marked unhealthy after {self._consecutive_failures} consecutive failures")
             return False
-        
+
         # Check if SSE task is still running
         if self.sse_task and self.sse_task.done():
             exception = self.sse_task.exception()
             if exception:
                 logger.warning(f"SSE task died: {exception}")
                 return False
-        
+
+        # FIXED: If we have a recent successful ping/tool call, we're healthy
+        if self._last_successful_ping and time.time() - self._last_successful_ping < 60.0:  # Success within last minute
+            return True
+
+        # FIXED: Default to healthy if no clear indicators of problems
+        logger.debug("No clear health indicators - defaulting to healthy")
         return True
 
-    async def get_tools(self) -> List[Dict[str, Any]]:
+    async def get_tools(self) -> list[dict[str, Any]]:
         """Get list of available tools from the server."""
         if not self._initialized:
             logger.error("Cannot get tools: transport not initialized")
             return []
-        
+
         start_time = time.time()
         try:
             response = await self._send_request("tools/list", {})
-            
-            if 'error' in response:
-                logger.error("Error getting tools: %s", response['error'])
+
+            if "error" in response:
+                logger.error("Error getting tools: %s", response["error"])
                 return []
-            
-            tools = response.get('result', {}).get('tools', [])
-            
+
+            tools = response.get("result", {}).get("tools", [])
+
             if self.enable_metrics:
                 response_time = time.time() - start_time
                 logger.debug("Retrieved %d tools in %.3fs", len(tools), response_time)
-            
+
             return tools
-            
+
         except Exception as e:
             logger.error("Error getting tools: %s", e)
             return []
 
-    async def call_tool(self, tool_name: str, arguments: Dict[str, Any], 
-                       timeout: Optional[float] = None) -> Dict[str, Any]:
+    async def call_tool(
+        self, tool_name: str, arguments: dict[str, Any], timeout: float | None = None
+    ) -> dict[str, Any]:
         """Execute a tool with the given arguments."""
         if not self._initialized:
-            return {
-                "isError": True,
-                "error": "Transport not initialized"
-            }
+            return {"isError": True, "error": "Transport not initialized"}
 
         start_time = time.time()
         if self.enable_metrics:
@@ -481,51 +493,37 @@ class SSETransport(MCPBaseTransport):
 
         try:
             logger.debug("Calling tool '%s' with arguments: %s", tool_name, arguments)
-            
+
             response = await self._send_request(
-                "tools/call",
-                {
-                    "name": tool_name,
-                    "arguments": arguments
-                },
-                timeout=timeout
+                "tools/call", {"name": tool_name, "arguments": arguments}, timeout=timeout
             )
-            
-            if 'error' in response:
+
+            if "error" in response:
                 if self.enable_metrics:
                     self._update_metrics(time.time() - start_time, False)
-                
-                return {
-                    "isError": True,
-                    "error": response['error'].get('message', 'Unknown error')
-                }
-            
+
+                return {"isError": True, "error": response["error"].get("message", "Unknown error")}
+
             # Extract and normalize result using base class method
-            result = response.get('result', {})
+            result = response.get("result", {})
             normalized_result = self._normalize_mcp_response({"result": result})
-            
+
             if self.enable_metrics:
                 self._update_metrics(time.time() - start_time, True)
-            
+
             return normalized_result
-            
-        except asyncio.TimeoutError:
+
+        except TimeoutError:
             if self.enable_metrics:
                 self._update_metrics(time.time() - start_time, False)
-            
-            return {
-                "isError": True,
-                "error": "Tool execution timed out"
-            }
+
+            return {"isError": True, "error": "Tool execution timed out"}
         except Exception as e:
             if self.enable_metrics:
                 self._update_metrics(time.time() - start_time, False)
-            
+
             logger.error("Error calling tool '%s': %s", tool_name, e)
-            return {
-                "isError": True,
-                "error": str(e)
-            }
+            return {"isError": True, "error": str(e)}
 
     def _update_metrics(self, response_time: float, success: bool) -> None:
         """Update performance metrics."""
@@ -533,39 +531,37 @@ class SSETransport(MCPBaseTransport):
             self._metrics["successful_calls"] += 1
         else:
             self._metrics["failed_calls"] += 1
-            
+
         self._metrics["total_time"] += response_time
         if self._metrics["total_calls"] > 0:
-            self._metrics["avg_response_time"] = (
-                self._metrics["total_time"] / self._metrics["total_calls"]
-            )
+            self._metrics["avg_response_time"] = self._metrics["total_time"] / self._metrics["total_calls"]
 
-    async def list_resources(self) -> Dict[str, Any]:
+    async def list_resources(self) -> dict[str, Any]:
         """List available resources from the server."""
         if not self._initialized:
             return {}
-        
+
         try:
             response = await self._send_request("resources/list", {}, timeout=10.0)
-            if 'error' in response:
-                logger.debug("Resources not supported: %s", response['error'])
+            if "error" in response:
+                logger.debug("Resources not supported: %s", response["error"])
                 return {}
-            return response.get('result', {})
+            return response.get("result", {})
         except Exception as e:
             logger.debug("Error listing resources: %s", e)
             return {}
 
-    async def list_prompts(self) -> Dict[str, Any]:
+    async def list_prompts(self) -> dict[str, Any]:
         """List available prompts from the server."""
         if not self._initialized:
             return {}
-        
+
         try:
             response = await self._send_request("prompts/list", {}, timeout=10.0)
-            if 'error' in response:
-                logger.debug("Prompts not supported: %s", response['error'])
+            if "error" in response:
+                logger.debug("Prompts not supported: %s", response["error"])
                 return {}
-            return response.get('result', {})
+            return response.get("result", {})
         except Exception as e:
             logger.debug("Error listing prompts: %s", e)
             return {}
@@ -574,16 +570,16 @@ class SSETransport(MCPBaseTransport):
         """Close the transport and clean up resources."""
         if not self._initialized:
             return
-        
+
         # Log final metrics
         if self.enable_metrics and self._metrics["total_calls"] > 0:
             logger.debug(
                 "SSE transport closing - Total calls: %d, Success rate: %.1f%%, Avg response time: %.3fs",
                 self._metrics["total_calls"],
                 (self._metrics["successful_calls"] / self._metrics["total_calls"] * 100),
-                self._metrics["avg_response_time"]
+                self._metrics["avg_response_time"],
             )
-        
+
         await self._cleanup()
 
     async def _cleanup(self) -> None:
@@ -591,30 +587,28 @@ class SSETransport(MCPBaseTransport):
         # Cancel SSE processing task
         if self.sse_task and not self.sse_task.done():
             self.sse_task.cancel()
-            try:
+            with contextlib.suppress(asyncio.CancelledError):
                 await self.sse_task
-            except asyncio.CancelledError:
-                pass
-        
+
         # Close SSE stream context
         if self.sse_stream_context:
             try:
                 await self.sse_stream_context.__aexit__(None, None, None)
             except Exception as e:
                 logger.debug("Error closing SSE stream: %s", e)
-        
+
         # Close HTTP clients
         if self.stream_client:
             await self.stream_client.aclose()
-        
+
         if self.send_client:
             await self.send_client.aclose()
-        
+
         # Cancel any pending requests
-        for request_id, future in self.pending_requests.items():
+        for _request_id, future in self.pending_requests.items():
             if not future.done():
                 future.cancel()
-        
+
         # Reset state
         self._initialized = False
         self.session_id = None
@@ -625,10 +619,30 @@ class SSETransport(MCPBaseTransport):
         self.sse_stream_context = None
         self.stream_client = None
         self.send_client = None
+        # FIXED: Reset health tracking
+        self._consecutive_failures = 0
+        self._last_successful_ping = None
+        self._initialization_time = None
 
-    def get_metrics(self) -> Dict[str, Any]:
-        """Get performance and connection metrics."""
-        return self._metrics.copy()
+    def get_metrics(self) -> dict[str, Any]:
+        """Get performance and connection metrics with health info."""
+        metrics = self._metrics.copy()
+        metrics.update(
+            {
+                "is_connected": self.is_connected(),
+                "consecutive_failures": self._consecutive_failures,
+                "max_consecutive_failures": self._max_consecutive_failures,
+                "last_successful_ping": self._last_successful_ping,
+                "initialization_time_timestamp": self._initialization_time,
+                "grace_period_active": (
+                    self._initialization_time
+                    and time.time() - self._initialization_time < self._connection_grace_period
+                )
+                if self._initialization_time
+                else False,
+            }
+        )
+        return metrics
 
     def reset_metrics(self) -> None:
         """Reset performance metrics."""
@@ -641,10 +655,10 @@ class SSETransport(MCPBaseTransport):
             "last_ping_time": self._metrics.get("last_ping_time"),
             "initialization_time": self._metrics.get("initialization_time"),
             "session_discoveries": self._metrics.get("session_discoveries", 0),
-            "stream_errors": 0
+            "stream_errors": 0,
         }
 
-    def get_streams(self) -> List[tuple]:
+    def get_streams(self) -> list[tuple]:
         """SSE transport doesn't expose raw streams."""
         return []
 
@@ -657,4 +671,4 @@ class SSETransport(MCPBaseTransport):
 
     async def __aexit__(self, exc_type, exc_val, exc_tb):
         """Context manager cleanup."""
-        await self.close()
+        await self.close()