Always use the failures-only reporter for testing (#160)

The default reporter has verbose output which is not useful but eats up
a lot of input tokens once it's in the history. Always run tests with
`--reporter=failures-only` to reduce the output size. This output format
still includes a count of passing tests.

Update changelog to focus on SDK version in the header and merge the
updates.
diff --git a/pkgs/dart_mcp_server/CHANGELOG.md b/pkgs/dart_mcp_server/CHANGELOG.md
index cdb9a04..fdffc6c 100644
--- a/pkgs/dart_mcp_server/CHANGELOG.md
+++ b/pkgs/dart_mcp_server/CHANGELOG.md
@@ -1,8 +1,4 @@
-# 0.1.1-wip
-
-* Instruct clients to prefer MCP tools over running tools in the shell.
-
-# 0.1.0 (Dart SDK 3.8.0)
+# Dart SDK 3.8.0 - WP
 
 * Handle relative paths under roots without trailing slashes.
 * Fix executable paths for dart/flutter on windows.
@@ -40,3 +36,5 @@
 * Add a hot reload tool.
 * Add basic analysis support.
 * Add the beginnings of a Dart tooling MCP server.
+* Instruct clients to prefer MCP tools over running tools in the shell.
+* Reduce output size of `run_tests` tool to save on input tokens.
diff --git a/pkgs/dart_mcp_server/lib/src/mixins/dash_cli.dart b/pkgs/dart_mcp_server/lib/src/mixins/dash_cli.dart
index be32e73..a6f8bfc 100644
--- a/pkgs/dart_mcp_server/lib/src/mixins/dash_cli.dart
+++ b/pkgs/dart_mcp_server/lib/src/mixins/dash_cli.dart
@@ -68,7 +68,7 @@
   Future<CallToolResult> _runTests(CallToolRequest request) async {
     return runCommandInRoots(
       request,
-      arguments: ['test'],
+      arguments: ['test', '--reporter=failures-only'],
       commandDescription: 'dart|flutter test',
       processManager: processManager,
       knownRoots: await roots,
diff --git a/pkgs/dart_mcp_server/pubspec.yaml b/pkgs/dart_mcp_server/pubspec.yaml
index aa20154..b560d11 100644
--- a/pkgs/dart_mcp_server/pubspec.yaml
+++ b/pkgs/dart_mcp_server/pubspec.yaml
@@ -1,5 +1,4 @@
 name: dart_mcp_server
-version: 0.1.1-wip
 description: >-
   An MCP server for Dart projects, exposing various developer tools to AI
   models.
diff --git a/pkgs/dart_mcp_server/test/tools/dart_cli_test.dart b/pkgs/dart_mcp_server/test/tools/dart_cli_test.dart
index 26697d6..d516ad5 100644
--- a/pkgs/dart_mcp_server/test/tools/dart_cli_test.dart
+++ b/pkgs/dart_mcp_server/test/tools/dart_cli_test.dart
@@ -166,13 +166,19 @@
           command: [
             endsWith(flutterExecutableName),
             'test',
+            '--reporter=failures-only',
             'foo_test.dart',
             'bar_test.dart',
           ],
           workingDirectory: exampleFlutterAppRoot.path,
         )),
         equalsCommand((
-          command: [endsWith(dartExecutableName), 'test', 'zip_test.dart'],
+          command: [
+            endsWith(dartExecutableName),
+            'test',
+            '--reporter=failures-only',
+            'zip_test.dart',
+          ],
           workingDirectory: dartCliAppRoot.path,
         )),
       ]);