chore(unity): 固定Unity CLI Loop本地包并移除UnitySkills依赖

Client/Packages/io.github.hatayama.uloopmcp/Editor/Api/McpTools/RunTests/Skill/SKILL.md

@@ -0,0 +1,71 @@
+---
+name: uloop-run-tests
+description: "Run Unity Test Runner and report detailed results. Use for EditMode/PlayMode tests, change verification, or failure diagnosis."
+---
+
+# uloop run-tests
+
+Execute Unity Test Runner. This command requires the Unity Test Framework package (`com.unity.test-framework`). If that package is not installed, the command returns `Success: false` with an unsupported message and does not affect the other Unity CLI Loop tools.
+
+When tests fail, NUnit XML results with error messages and stack traces are automatically saved. Read the XML file at `XmlPath` for detailed failure diagnosis.
+
+Before executing tests, `uloop run-tests` checks for unsaved loaded Scene changes and unsaved current Prefab Stage changes. If any are found, it returns `Success: false`, keeps `TestCount` at `0`, lists the unsaved items in `Message`, and does not start the Unity Test Runner. Save or discard those editor changes, then rerun the command. Use `--save-before-run true` only when the user explicitly asks to save editor changes before continuing.
+
+## Usage
+
+```bash
+uloop run-tests [options]
+```
+
+## Parameters
+
+| Parameter | Type | Default | Description |
+|-----------|------|---------|-------------|
+| `--test-mode` | string | `EditMode` | Test mode: `EditMode`, `PlayMode` |
+| `--filter-type` | string | `all` | Filter type: `all`, `exact`, `regex`, `assembly` |
+| `--filter-value` | string | - | Filter value (test name, pattern, or assembly) |
+| `--save-before-run` | boolean | `false` | Save unsaved loaded Scene changes and current Prefab Stage changes before running tests |
+
+## Global Options
+
+| Option | Description |
+|--------|-------------|
+| `--project-path <path>` | Optional. Use only when the target Unity project is not the current directory. |
+
+## Examples
+
+```bash
+# Run all EditMode tests
+uloop run-tests
+
+# Run PlayMode tests
+uloop run-tests --test-mode PlayMode
+
+# Save explicitly approved editor changes before running tests
+uloop run-tests --save-before-run true
+
+# Run specific test
+uloop run-tests --filter-type exact --filter-value "MyTest.TestMethod"
+
+# Run tests matching pattern
+uloop run-tests --filter-type regex --filter-value ".*Integration.*"
+```
+
+## Output
+
+Returns JSON with:
+- `Success` (boolean): Whether all tests passed
+- `Message` (string): Summary message
+- `CompletedAt` (string): ISO timestamp when the run finished
+- `TestCount` (number): Total tests executed
+- `PassedCount` (number): Passed tests
+- `FailedCount` (number): Failed tests
+- `SkippedCount` (number): Skipped tests
+- `XmlPath` (string | null): Path to NUnit XML result file. `null` when no XML was saved; populated only when tests failed and the XML file exists on disk.
+
+### XML Result File
+
+When tests fail, NUnit XML results are automatically saved to `{project_root}/.uloop/outputs/TestResults/<timestamp>.xml`. The XML contains per-test-case results including:
+- Test name and full name
+- Pass/fail/skip status and duration
+- For failed tests: `<message>` (assertion error) and `<stack-trace>`

Client/Packages/io.github.hatayama.uloopmcp/Editor/Api/McpTools/RunTests/Skill/SKILL.md.meta

@@ -0,0 +1,7 @@
+fileFormatVersion: 2
+guid: cae04d651537c45d6817556ccd093c39
+TextScriptImporter:
+  externalObjects: {}
+  userData: 
+  assetBundleName: 
+  assetBundleVariant: