Update McpErrorCode XML docs and fix error code usage to align with MCP spec

docs/concepts/tasks/tasks.md

@@ -424,8 +424,8 @@ Task operations may throw <xref:ModelContextProtocol.McpException> with these er
 
 | Error Code | Scenario |
 |------------|----------|
-| `InvalidParams` | Invalid or nonexistent task ID |
-| `InvalidRequest` | Tool with `taskSupport: forbidden` called with task metadata, or tool with `taskSupport: required` called without task metadata |
+| `InvalidParams` | Invalid or nonexistent task ID or invalid cursor |
+| `InvalidParams` | Tool with `taskSupport: forbidden` called with task metadata, or tool with `taskSupport: required` called without task metadata |
 | `InternalError` | Task execution failure or result unavailable |
 
 Example error handling:

src/ModelContextProtocol.Core/McpErrorCode.cs

@@ -39,10 +39,19 @@ public enum McpErrorCode
     InvalidRequest = -32600,
 
     /// <summary>
-    /// Indicates that the requested method does not exist or is not available on the server.
+    /// Indicates that the requested method does not exist or is not available.
     /// </summary>
     /// <remarks>
-    /// This error is returned when the method name specified in the request cannot be found.
+    /// <para>
+    /// In MCP, this error is returned when a request is made for a method that requires a capability
+    /// that has not been declared. This can occur in either direction:
+    /// </para>
+    /// <list type="bullet">
+    /// <item><description>A server returning this error when the client requests a capability it doesn't support
+    /// (for example, requesting completions when the <c>completions</c> capability was not advertised).</description></item>
+    /// <item><description>A client returning this error when the server requests a capability it doesn't support
+    /// (for example, requesting roots when the client did not declare the <c>roots</c> capability).</description></item>
+    /// </list>
     /// </remarks>
     MethodNotFound = -32601,
 
@@ -51,15 +60,19 @@ public enum McpErrorCode
     /// </summary>
     /// <remarks>
     /// <para>
-    /// This error is returned for protocol-level parameter issues, such as:
+    /// In MCP, this error is returned for protocol-level parameter validation failures in various contexts:
     /// </para>
     /// <list type="bullet">
-    /// <item><description>Malformed requests that fail to satisfy the request schema (for example, CallToolRequest)</description></item>
-    /// <item><description>Unknown or unrecognized primitive names (for example, tool, prompt, or resource names)</description></item>
-    /// <item><description>Missing required protocol-level parameters</description></item>
+    /// <item><description><b>Tools</b>: Unknown tool name or invalid protocol-level tool arguments.</description></item>
+    /// <item><description><b>Prompts</b>: Unknown prompt name or missing required protocol-level arguments.</description></item>
+    /// <item><description><b>Pagination</b>: Invalid or expired cursor values.</description></item>
+    /// <item><description><b>Logging</b>: Invalid log level.</description></item>
+    /// <item><description><b>Tasks</b>: Invalid or nonexistent task ID or invalid cursor.</description></item>
+    /// <item><description><b>Elicitation</b>: Server requests an elicitation mode not declared in client capabilities.</description></item>
+    /// <item><description><b>Sampling</b>: Missing tool result or tool results mixed with other content.</description></item>
     /// </list>
     /// <para>
-    /// Note: Input validation errors within tool/prompt/resource arguments should be reported as execution errors
+    /// Note: Application-layer validation errors within tool/prompt/resource arguments should be reported as execution errors
     /// (for example, via <see cref="Protocol.CallToolResult.IsError"/>) rather than as protocol errors, allowing language
     /// models to receive error feedback and self-correct.
     /// </para>

src/ModelContextProtocol.Core/Server/McpServerImpl.cs

@@ -591,7 +591,7 @@ await originalListToolsHandler(request, cancellationToken).ConfigureAwait(false)
                         {
                             throw new McpProtocolException(
                                 $"Tool '{tool.ProtocolTool.Name}' does not support task-augmented execution.",
-                                McpErrorCode.MethodNotFound);
+                                McpErrorCode.InvalidParams);
                         }
 
                         // Task augmentation requested - return CreateTaskResult
@@ -604,7 +604,7 @@ await originalListToolsHandler(request, cancellationToken).ConfigureAwait(false)
                         throw new McpProtocolException(
                             $"Tool '{tool.ProtocolTool.Name}' requires task-augmented execution. " +
                             "Include a 'task' parameter with the request.",
-                            McpErrorCode.MethodNotFound);
+                            McpErrorCode.InvalidParams);
                     }
 
                     // Normal synchronous execution

tests/ModelContextProtocol.Tests/Server/McpServerTaskAugmentedValidationTests.cs

@@ -100,7 +100,7 @@ await client.CallToolAsync(
                 TestContext.Current.CancellationToken));
 
         Assert.Contains("does not support task-augmented execution", exception.Message, StringComparison.OrdinalIgnoreCase);
-        Assert.Equal(McpErrorCode.MethodNotFound, exception.ErrorCode);
+        Assert.Equal(McpErrorCode.InvalidParams, exception.ErrorCode);
     }
 
     [Fact]
@@ -227,7 +227,7 @@ await client.CallToolAsync(
                 TestContext.Current.CancellationToken));
 
         Assert.Contains("requires task-augmented execution", exception.Message, StringComparison.OrdinalIgnoreCase);
-        Assert.Equal(McpErrorCode.MethodNotFound, exception.ErrorCode);
+        Assert.Equal(McpErrorCode.InvalidParams, exception.ErrorCode);
     }
 
     [Fact]

tests/ModelContextProtocol.Tests/Server/TaskCancellationIntegrationTests.cs

@@ -460,7 +460,7 @@ public async Task CompletedTask_CannotTransitionToOtherStatus()
 
         Assert.Equal(McpTaskStatus.Completed, taskStatus.Status);
 
-        // Act - Try to cancel a completed task
+        // Act - Try to cancel a completed task (should be idempotent)
         var cancelResult = await client.CancelTaskAsync(taskId, cancellationToken: TestContext.Current.CancellationToken);
 
         // Assert - Status should still be completed (not cancelled)
@@ -500,7 +500,7 @@ public async Task FailedTask_CannotTransitionToOtherStatus()
 
         Assert.Equal(McpTaskStatus.Failed, taskStatus.Status);
 
-        // Act - Try to cancel a failed task
+        // Act - Try to cancel a failed task (should be idempotent)
         var cancelResult = await client.CancelTaskAsync(taskId, cancellationToken: TestContext.Current.CancellationToken);
 
         // Assert - Status should still be failed

tests/ModelContextProtocol.Tests/Server/ToolTaskSupportTests.cs

@@ -526,8 +526,8 @@ public async Task SyncTool_WithRequiredTaskSupport_CannotBeCalledDirectly()
                 arguments: new Dictionary<string, object?> { ["input"] = "test" },
                 cancellationToken: TestContext.Current.CancellationToken).AsTask());
 
-        // The server returns MethodNotFound because direct invocation is not allowed for required-task tools
-        Assert.Equal(McpErrorCode.MethodNotFound, exception.ErrorCode);
+        // The server returns InvalidParams because direct invocation is not allowed for required-task tools
+        Assert.Equal(McpErrorCode.InvalidParams, exception.ErrorCode);
         Assert.Contains("task", exception.Message, StringComparison.OrdinalIgnoreCase);
     }