feat: introduce MCP server, cross-platform runtime, and Showcase UI reorganization

- Showcase UI Reorganization:
  * Move web UI dashboard from root directory into apps/showcase_ui/ for cleaner project structure
  * Add 'artemis ui' CLI command (artemis/interfaces/cli/commands/ui.py) and launcher scripts for the interactive showcase console
- Standalone MCP Server & Notifiers (mcp_server/):
  * Implement dedicated MCP server package with mobile automation tools (device_state, inspect_trace, task_manager, task_runner)
  * Add multi-channel notifiers (AgentAPI, Desktop notifications, Webhooks, File output, and Composite routing)
  * Introduce background task runner and trace store for autonomous mobile execution monitoring
- Cross-Platform Runtime & Toolchain Support:
  * Add artemis/platform/ with OS-specific implementations for Linux, macOS (Darwin), and Windows
  * Add artemis/runtime/supervisor.py for resilient process and agent runtime supervision
  * Add artemis/toolchain/ with dependency resolution and SDK descriptors
  * Add cross-platform launcher scripts (start.sh, start.bat) and installer scripts (scripts/install_deps.*)
- Core Enhancements & Diagnostics:
  * Enhance coordinate normalization, OCR API fallback, and perception pipeline
  * Add credential validation (credentials_validator.py) and system diagnostics
  * Update bilingual documentation (README.md, README_CN.md) and architecture diagrams
- Test Suite Expansion:
  * Add integration tests for cloud gateway, remote ADB, and Artemis login
  * Add unit tests covering core diagnostics, platform detection, toolchain, credentials validator, optional OCR, and MCP tools

artemis/graph/graph.py

@@ -358,6 +358,61 @@ class CyFunctionDetector(metaclass=_CyFunctionDetectorMeta):
     pass
 
 
+def check_plan_mutation_rejections(
+    content_before: str, content_after: str, state: State | None = None
+) -> str | None:
+    """Checks if a task plan modification violates system safety or loop integrity rules."""
+    lines_after = content_after.splitlines()
+    top_level_after = [line for line in lines_after if line.startswith("- [")]
+
+    # 1. Incomplete nested subgoals check
+    if top_level_after and all(line.startswith("- [x]") for line in top_level_after):
+        has_incomplete = any(re.match(r"^\s*-\s*\[[ /]\]", line) for line in lines_after)
+        if has_incomplete:
+            return (
+                "Your changes to the task plan were not applied. Your changes would cause the"
+                " task to end completely, but there seem to be goals in the task plan"
+                " that have not been marked as completed. If you believe these goals"
+                " are already completed, please update their statuses."
+            )
+
+    # 2. Continuous monitoring / Loop milestone protection check
+    lines_before = content_before.splitlines()
+    top_level_before = [line for line in lines_before if line.startswith("- [")]
+
+    continuous_pattern = re.compile(r"\[Loop\].*?(?:continuous|until\s*manual)", re.IGNORECASE)
+    has_continuous_before = any(continuous_pattern.search(line) for line in top_level_before)
+
+    injected = getattr(state, "injected_instruction", None) if state else None
+    user_stopped = bool(
+        injected
+        and any(
+            w in str(injected).lower()
+            for w in ["stop", "finish", "end", "停止", "结束", "退出", "quit"]
+        )
+    )
+
+    if has_continuous_before and not user_stopped:
+        has_continuous_after = any(continuous_pattern.search(line) for line in top_level_after)
+        if not has_continuous_after:
+            return (
+                "Your changes to the task plan were rejected. You cannot delete an active [Loop] "
+                "continuous monitoring milestone or remove its [Loop] tag. Please keep the [Loop] "
+                "milestone active (in progress [/]) and record each check cycle as an indented "
+                "subtask."
+            )
+        for line in top_level_after:
+            if continuous_pattern.search(line) and line.startswith("- [x]"):
+                return (
+                    "Your changes to the task plan were rejected. Ongoing continuous monitoring "
+                    "milestones (exit condition: 'until manually stopped') must remain in progress "
+                    "([/]) and cannot be unilaterally marked as completed [x]. Please keep it "
+                    "active ([/]) and record each polling check as an indented subtask."
+                )
+
+    return None
+
+
 class NoteArgs(BaseModel):
     model_config = {"ignored_types": (CyFunctionDetector,)}
     key: str = Field(..., description=SAVE_NOTE_ARG_KEY_DESC)
@@ -392,36 +447,15 @@ def wrap_note_tool(ctx: ArtemisContext, original_tool):
             if task_plan_path.exists():
                 content_after = task_plan_path.read_text(encoding="utf-8")
 
-            lines = content_after.splitlines()
-            top_level_lines = [line for line in lines if line.startswith("- [")]
-            if top_level_lines and all(line.startswith("- [x]") for line in top_level_lines):
-                has_incomplete = any(re.match(r"^\s*-\s*\[[ /]\]", line) for line in lines)
-                if has_incomplete:
-                    logger.info(
-                        "Rejected: Top-level subgoals are complete but"
-                        " incomplete nested subgoals remain."
-                    )
-                    if task_plan_path.exists():
-                        task_plan_path.write_text(content_before, encoding="utf-8")
+            rejection_message = check_plan_mutation_rejections(content_before, content_after, state)
+            if rejection_message:
+                logger.info(f"Rejected task plan update: {rejection_message}")
+                if task_plan_path.exists():
+                    task_plan_path.write_text(content_before, encoding="utf-8")
 
-                    return Command(
-                        update={
-                            VALIDATOR_MESSAGES_KEY: [
-                                SystemMessage(
-                                    content=(
-                                        "Your changes to the task plan were not"
-                                        " applied. Your changes would cause the"
-                                        " task to end completely, but there"
-                                        " seem to be goals in the task plan"
-                                        " that have not been marked as"
-                                        " completed. If you believe these goals"
-                                        " are already completed, please update"
-                                        " their statuses."
-                                    )
-                                )
-                            ]
-                        }
-                    )
+                return Command(
+                    update={VALIDATOR_MESSAGES_KEY: [SystemMessage(content=rejection_message)]}
+                )
 
             threshold = (
                 ctx.execution_setup.planner_validation_threshold
@@ -558,36 +592,15 @@ def wrap_update_note_tool(ctx: ArtemisContext, original_tool):
             if task_plan_path.exists():
                 content_after = task_plan_path.read_text(encoding="utf-8")
 
-            lines = content_after.splitlines()
-            top_level_lines = [line for line in lines if line.startswith("- [")]
-            if top_level_lines and all(line.startswith("- [x]") for line in top_level_lines):
-                has_incomplete = any(re.match(r"^\s*-\s*\[[ /]\]", line) for line in lines)
-                if has_incomplete:
-                    logger.info(
-                        "Rejected: Top-level subgoals are complete but"
-                        " incomplete nested subgoals remain."
-                    )
-                    if task_plan_path.exists():
-                        task_plan_path.write_text(content_before, encoding="utf-8")
+            rejection_message = check_plan_mutation_rejections(content_before, content_after, state)
+            if rejection_message:
+                logger.info(f"Rejected task plan update via update_note: {rejection_message}")
+                if task_plan_path.exists():
+                    task_plan_path.write_text(content_before, encoding="utf-8")
 
-                    return Command(
-                        update={
-                            VALIDATOR_MESSAGES_KEY: [
-                                SystemMessage(
-                                    content=(
-                                        "Your changes to the task plan were not"
-                                        " applied. Your changes would cause the"
-                                        " task to end completely, but there"
-                                        " seem to be goals in the task plan"
-                                        " that have not been marked as"
-                                        " completed. If you believe these goals"
-                                        " are already completed, please update"
-                                        " their statuses."
-                                    )
-                                )
-                            ]
-                        }
-                    )
+                return Command(
+                    update={VALIDATOR_MESSAGES_KEY: [SystemMessage(content=rejection_message)]}
+                )
 
             threshold = (
                 ctx.execution_setup.planner_validation_threshold
@@ -710,6 +723,18 @@ async def get_graph(ctx: ArtemisContext) -> CompiledStateGraph:
 
         lines = task_plan_content.splitlines()
 
+        # Check for continuous monitoring loop that should not terminate prematurely
+        continuous_pattern = re.compile(r"\[Loop\].*?(?:continuous|until\s*manual)", re.IGNORECASE)
+        has_continuous_monitoring = any(continuous_pattern.search(line) for line in lines)
+        injected = getattr(state, "injected_instruction", None) if state else None
+        user_stopped = bool(
+            injected
+            and any(
+                w in str(injected).lower()
+                for w in ["stop", "finish", "end", "停止", "结束", "退出", "quit"]
+            )
+        )
+
         # Check for all completed
         top_level_lines = [line for line in lines if line.startswith("- [")]
 
@@ -719,6 +744,12 @@ async def get_graph(ctx: ArtemisContext) -> CompiledStateGraph:
 
         all_done = all(line.startswith("- [x]") for line in top_level_lines)
         if all_done:
+            if has_continuous_monitoring and not user_stopped:
+                logger.warning(
+                    "All subgoals marked [x] but task plan has active continuous monitoring"
+                    " without user stop signal. Continuing."
+                )
+                return "continue"
             logger.info("All subgoals are completed, ending the goal")
             return "end"
 

artemis/graph/perception.py

@@ -23,7 +23,7 @@ from artemis.controllers.unified_controller import UnifiedMobileController
 from artemis.data_engine.trace import trace
 from artemis.graph.state import State
 from artemis.utils.logger import get_logger
-from artemis.utils.ocr_api import perform_ocr
+from artemis.utils.ocr_api import is_ocr_configured, perform_ocr
 from artemis.utils.ocr_xml_fusion import (
     _crop_image_remove_status_bar,
     _detect_status_bar_height,
@@ -124,25 +124,35 @@ async def perception_node(state: State, ctx: ArtemisContext) -> dict:
         ctx.device.device_height = device_data.height
         logger.info(f"Updated context device dimensions: {device_data.width}x{device_data.height}")
 
-    # 3. Perform OCR
-    screen_height = device_data.height
-    status_bar_height = _detect_status_bar_height(xml_hierarchy, screen_height)
-    if status_bar_height > 0:
-        logger.info(
-            f"Detected status bar height: {status_bar_height}. Cropping screenshot for OCR."
-        )
-        cropped_b64, _, _ = await asyncio.to_thread(
-            _crop_image_remove_status_bar,
-            latest_screenshot_b64,
-            status_bar_height,
-        )
-        raw_ocr_results = await perform_ocr(cropped_b64)
-        ocr_results = _map_coordinates_back(raw_ocr_results, status_bar_height)
+    # 3. Perform OCR (if configured, else fallback to pure XML layout)
+    ocr_results = []
+    if is_ocr_configured():
+        screen_height = device_data.height
+        status_bar_height = _detect_status_bar_height(xml_hierarchy, screen_height)
+        try:
+            if status_bar_height > 0:
+                logger.info(
+                    f"Detected status bar height: {status_bar_height}. Cropping screenshot for OCR."
+                )
+                cropped_b64, _, _ = await asyncio.to_thread(
+                    _crop_image_remove_status_bar,
+                    latest_screenshot_b64,
+                    status_bar_height,
+                )
+                raw_ocr_results = await perform_ocr(cropped_b64)
+                ocr_results = _map_coordinates_back(raw_ocr_results, status_bar_height)
+            else:
+                logger.info("No status bar detected. Performing OCR on full screenshot.")
+                ocr_results = await perform_ocr(latest_screenshot_b64)
+        except Exception as ocr_err:
+            logger.warning(
+                f"Perception OCR execution encountered error: {ocr_err}. Proceeding with XML-only layout."
+            )
+            ocr_results = []
     else:
-        logger.info("No status bar detected. Performing OCR on full screenshot.")
-        ocr_results = await perform_ocr(latest_screenshot_b64)
+        logger.debug("OCR is not configured. Proceeding with XML-only layout.")
 
-    # 4. Fuse OCR with XML
+    # 4. Fuse OCR with XML (if ocr_results is empty, returns original xml_hierarchy intact)
     fused_xml = fuse_ocr_with_xml(xml_hierarchy, ocr_results)
 
     # 5. Save to Data Engine in background thread