mcp-instana 0.3.1__py3-none-any.whl → 0.7.0__py3-none-any.whl

src/infrastructure/infrastructure_analyze_new.py

@@ -0,0 +1,600 @@
+"""
+Infrastructure Analyze Tool - Option 2 Architecture
+
+This tool implements the complete Option 2 flow:
+1. LLM provides high-level intent (entity, metric, filters)
+2. Server normalizes intent
+3. Server resolves exact constants from schemas
+4. Server handles elicitation if ambiguous
+5. Server compiles complete payload
+6. Server calls Instana API
+
+Key benefit: LLM never sees schema complexity, reducing tokens by 99.4%
+"""
+
+import logging
+from pathlib import Path
+from typing import Any, Dict, List, Optional
+
+try:
+    from instana_client.api.infrastructure_analyze_api import InfrastructureAnalyzeApi
+except ImportError as e:
+    logger = logging.getLogger(__name__)
+    logger.error(f"Error importing Instana SDK: {e}", exc_info=True)
+    raise
+
+from mcp.types import EmbeddedResource, TextContent, ToolAnnotations
+
+from src.core.utils import BaseInstanaClient, register_as_tool, with_header_auth
+from src.infrastructure.elicitation_handler import ElicitationHandler
+from src.infrastructure.entity_registry import EntityCapabilityRegistry
+
+logger = logging.getLogger(__name__)
+
+
+class InfrastructureAnalyzeOption2(BaseInstanaClient):
+    """
+    Infrastructure analyze tool using Option 2 architecture.
+
+    This tool demonstrates the complete server-side payload assembly approach:
+    - LLM provides simple intent (~300 tokens)
+    - Server resolves all constants from schemas (~48,000 tokens saved)
+    - Zero hallucination on metric/tag names
+    """
+
+    def __init__(self, read_token: str, base_url: str, schema_dir: Optional[Path] = None):
+        """
+        Initialize the Infrastructure Analyze Option 2 tool.
+
+        Args:
+            read_token: Instana API read token
+            base_url: Instana API base URL
+            schema_dir: Path to schema directory (defaults to ../schema)
+        """
+        super().__init__(read_token=read_token, base_url=base_url)
+
+        # Initialize Option 2 components
+        if schema_dir is None:
+            # Default to schema directory relative to this file
+            schema_dir = Path(__file__).parent.parent.parent / "schema"
+
+        self.registry = EntityCapabilityRegistry(schema_dir)
+        self.elicitation_handler = ElicitationHandler()
+
+        logger.info(f"Initialized Option 2 tool with {len(self.registry.get_entity_types())} entity types")
+
+    @register_as_tool(
+        title="Analyze Infrastructure with Elicitation (Two-Pass)",
+        annotations=ToolAnnotations(readOnlyHint=True, destructiveHint=False)
+    )
+    @with_header_auth(InfrastructureAnalyzeApi)
+    async def analyze_infrastructure_elicitation(
+        self,
+        intent: Optional[str] = None,
+        entity: Optional[str] = None,
+        selections: Optional[Dict[str, Any]] = None,
+        ctx=None,
+        api_client=None
+    ) -> List[Any]:
+        """
+        Two-pass infrastructure analysis using machine-facing elicitation.
+
+        **Pass 1 - Intent to Schema:**
+        Provide intent and entity hint. Server returns full schema via MCP elicitation.
+
+        Parameters:
+        - intent: Natural language query (e.g., "maximum heap size of JVM on host galactica1")
+        - entity: Entity hint (e.g., "jvm", "kubernetes", "docker")
+
+        **Pass 2 - Selections to Results:**
+        Provide exact selections from schema. Server builds payload and calls API.
+
+        Parameters:
+        - selections: Dict with:
+          - entity_type: Exact entity type from schema (e.g., "jvmRuntimePlatform")
+          - metrics: Array of exact metric names from schema (e.g., ["jvm.heap.maxSize", "jvm.heap.used"])
+          - aggregation: Aggregation type (e.g., "max", "mean", "sum")
+          - filters: List of dicts with name/value pairs (e.g., [{"name": "host.name", "value": "galactica1"}])
+          - groupBy: (optional) Array of tag names to group entities by (e.g., ["host.name"], ["kubernetes.namespace.name"])
+          - timeRange: (optional) Time range string (e.g., "1h", "30m", "2h", "1d"). Default: "1h"
+          - order: (optional) Dict with "by" (metric name) and "direction" ("ASC" or "DESC")
+
+        Returns:
+            List of MCP content blocks (TextContent or EmbeddedResource)
+        """
+        try:
+            # Route based on input
+            if intent is not None and entity is not None:
+                # Pass 1: Intent → Elicitation
+                return await self._handle_pass1_intent({"intent": intent, "entity": entity})
+            elif selections is not None:
+                # Pass 2: Selections → API Call
+                return await self._handle_pass2_selections({"selections": selections}, api_client)
+            else:
+                return [TextContent(
+                    type="text",
+                    text="Error: Invalid input. Provide either (intent + entity) for Pass 1, or (selections) for Pass 2."
+                )]
+        except Exception as e:
+            logger.error(f"Error in analyze_infrastructure_elicitation: {e}", exc_info=True)
+            return [TextContent(
+                type="text",
+                text=f"Error: {e!s}"
+            )]
+
+    async def _handle_pass1_intent(self, arguments: Dict[str, Any]) -> List[Any]:
+        """
+        Handle Pass 1: Intent → Schema Elicitation
+
+        Args:
+            arguments: Dict with 'intent' and 'entity' keys
+
+        Returns:
+            List of MCP content blocks with schema
+        """
+        intent = arguments.get("intent", "")
+        entity_hint = arguments.get("entity", "").lower()
+        intent_lower = intent.lower()
+
+        logger.info(f"Pass 1: intent='{intent}', entity_hint='{entity_hint}'")
+
+        # Entity type resolution with priority (more specific first)
+        # Check for specific matches first, then fall back to general ones
+        entity_type = None
+
+        # Priority 1: Specific multi-word matches
+        if "kubernetes deployment" in entity_hint or "k8s deployment" in entity_hint:
+            entity_type = "kubernetesDeployment"
+        elif "kubernetes pod" in entity_hint or "k8s pod" in entity_hint:
+            entity_type = "kubernetesPod"
+        elif "docker container" in entity_hint:
+            entity_type = "dockerContainer"
+        elif "ibm mq" in entity_hint or "ibmmq" in entity_hint:
+            entity_type = "ibmMqQueue"
+        elif "db2 database" in entity_hint:
+            entity_type = "db2Database"
+        # Priority 2: Single word specific matches
+        elif "deployment" in entity_hint:
+            entity_type = "kubernetesDeployment"
+        elif "pod" in entity_hint:
+            entity_type = "kubernetesPod"
+        elif "jvm" in entity_hint or "java" in entity_hint:
+            entity_type = "jvmRuntimePlatform"
+        elif "docker" in entity_hint or "container" in entity_hint:
+            entity_type = "dockerContainer"
+        elif "mq" in entity_hint or "queue" in entity_hint:
+            entity_type = "ibmMqQueue"
+        elif "db2" in entity_hint or "database" in entity_hint:
+            entity_type = "db2Database"
+        elif "host" in entity_hint or "server" in entity_hint or "machine" in entity_hint:
+            entity_type = "host"
+        # Priority 3: Generic kubernetes - use intent context to disambiguate
+        elif "kubernetes" in entity_hint or "k8s" in entity_hint:
+            # Smart disambiguation based on intent keywords
+            deployment_keywords = ["deployment", "replica", "availabletodesiredreplica", "desiredreplica", "availablereplica"]
+            pod_keywords = ["pod", "restart", "container"]
+
+            # Check intent for deployment-specific keywords
+            if any(keyword in intent_lower for keyword in deployment_keywords):
+                entity_type = "kubernetesDeployment"
+                logger.info("Resolved ambiguous 'kubernetes' to 'kubernetesDeployment' based on intent keywords")
+            # Check intent for pod-specific keywords
+            elif any(keyword in intent_lower for keyword in pod_keywords):
+                entity_type = "kubernetesPod"
+                logger.info("Resolved ambiguous 'kubernetes' to 'kubernetesPod' based on intent keywords")
+            else:
+                # Still ambiguous - ask for clarification
+                return [TextContent(
+                    type="text",
+                    text="Ambiguous entity type 'kubernetes'. Please specify: 'kubernetes pod' or 'kubernetes deployment'"
+                )]
+
+        if not entity_type:
+            return [TextContent(
+                type="text",
+                text=f"Error: Unknown entity type '{entity_hint}'. Supported: jvm, kubernetes pod, kubernetes deployment, docker, ibmmq, db2"
+            )]
+
+        logger.info(f"Resolved entity type: {entity_type}")
+
+        # Load full schema
+        schema = self.registry.get_full_schema(entity_type)
+        if not schema:
+            return [TextContent(
+                type="text",
+                text=f"Error: Could not load schema for {entity_type}"
+            )]
+
+        logger.info(f"Loaded schema for {entity_type}")
+
+        # Create elicitation response
+        elicitation_content = self.elicitation_handler.create_schema_elicitation(
+            entity_type=entity_type,
+            schema=schema,
+            intent=intent
+        )
+
+        logger.info(f"Created elicitation with {len(elicitation_content)} content blocks")
+
+        return elicitation_content
+
+    async def _handle_pass2_selections(
+        self,
+        arguments: Dict[str, Any],
+        api_client
+    ) -> List[Any]:
+        """
+        Handle Pass 2: Selections → API Call
+
+        Args:
+            arguments: Dict with 'selections' key containing user selections
+            api_client: Instana API client
+
+        Returns:
+            List with TextContent containing formatted results
+        """
+        selections = arguments.get("selections", {})
+
+        entity_type = selections.get("entity_type")
+        # Support both "metrics" (correct) and "metric" (backward compatibility)
+        metrics = selections.get("metrics") or selections.get("metric", [])
+        aggregation = selections.get("aggregation", "mean")
+        filters = selections.get("filters", [])
+        group_by = selections.get("groupBy", [])
+        order = selections.get("order")  # Optional: {"by": "metric_name", "direction": "ASC|DESC"}
+        time_range = selections.get("timeRange", "1h")  # Default to 1 hour
+        pagination = selections.get("pagination", {})  # Optional: {"page": 1, "pageSize": 20} or {"offset": 0, "limit": 20}
+
+        # Ensure metrics is a list
+        if not isinstance(metrics, list):
+            metrics = [metrics] if metrics else []
+
+        # Ensure groupBy is a list
+        if not isinstance(group_by, list):
+            group_by = [group_by] if group_by else []
+
+        logger.info(f"Pass 2: entity_type={entity_type}, metrics={metrics}, aggregation={aggregation}, groupBy={group_by}, order={order}, timeRange={time_range}")
+
+        if not entity_type or not metrics:
+            return [TextContent(
+                type="text",
+                text="Error: selections must include 'entity_type' and 'metrics' (array)"
+            )]
+
+        # Validate groupBy if provided
+        if group_by and len(group_by) > 5:
+            return [TextContent(
+                type="text",
+                text="Error: groupBy can have maximum 5 tag names"
+            )]
+
+        # Build payload using payload compiler
+        # For PoC, we'll build a simple payload directly
+        from instana_client.models.cursor_pagination import CursorPagination
+        from instana_client.models.get_infrastructure_groups_query import (
+            GetInfrastructureGroupsQuery,
+        )
+        from instana_client.models.get_infrastructure_query import (
+            GetInfrastructureQuery,
+        )
+        from instana_client.models.infra_metric_configuration import (
+            InfraMetricConfiguration,
+        )
+        from instana_client.models.order import Order
+        from instana_client.models.simple_metric_configuration import (
+            SimpleMetricConfiguration,
+        )
+        from instana_client.models.tag_filter import TagFilter
+        from instana_client.models.tag_filter_all_of_value import TagFilterAllOfValue
+        from instana_client.models.tag_filter_expression import TagFilterExpression
+        from instana_client.models.time_frame import TimeFrame
+
+        # Parse time range - supports relative, absolute timestamps, and human-readable dates
+        def parse_time_range(time_input):
+            """
+            Convert time range to TimeFrame parameters.
+
+            Supports three formats:
+            1. Relative: "1h", "30m", "2h", "1d" -> returns (window_size_ms, None)
+            2. Absolute (timestamps): {"from": timestamp_ms, "to": timestamp_ms} -> returns (window_size_ms, to_ms)
+            3. Absolute (date strings): {"from": "2026-01-24 12:25:00", "to": "2026-01-24 14:40:00"} -> returns (window_size_ms, to_ms)
+
+            Returns:
+                tuple: (window_size_ms, to_timestamp_ms or None)
+            """
+            from datetime import datetime
+
+            # Handle dict format for absolute time range
+            if isinstance(time_input, dict):
+                from_val = time_input.get("from")
+                to_val = time_input.get("to")
+
+                if from_val is not None and to_val is not None:
+                    # Check if values are strings (date format) or integers (timestamps)
+                    if isinstance(from_val, str) and isinstance(to_val, str):
+                        # Parse date strings to timestamps
+                        try:
+                            from datetime import timezone
+
+                            # Support multiple date formats
+                            date_formats = [
+                                "%Y-%m-%d %H:%M:%S",
+                                "%Y-%m-%d %H:%M",
+                                "%Y-%m-%dT%H:%M:%S",
+                                "%Y-%m-%dT%H:%M:%SZ",
+                                "%d-%B-%Y %H:%M",  # e.g., "24-January-2026 12:25"
+                                "%d-%b-%Y %H:%M",   # e.g., "24-Jan-2026 12:25"
+                            ]
+
+                            from_dt = None
+                            to_dt = None
+
+                            for fmt in date_formats:
+                                try:
+                                    from_dt = datetime.strptime(from_val, fmt)
+                                    to_dt = datetime.strptime(to_val, fmt)
+                                    break
+                                except ValueError:
+                                    continue
+
+                            if from_dt is None or to_dt is None:
+                                logger.error(f"Could not parse date strings: from='{from_val}', to='{to_val}'")
+                                return (3600000, None)
+
+                            # Use system's local timezone for timestamp conversion
+                            from_ts = int(from_dt.timestamp() * 1000)
+                            to_ts = int(to_dt.timestamp() * 1000)
+                            window_size_ms = to_ts - from_ts
+
+                            logger.info(f"Parsed date strings (local timezone): from='{from_val}' ({from_ts}), to='{to_val}' ({to_ts}), window={window_size_ms}ms")
+                            return (window_size_ms, to_ts)
+
+                        except Exception as e:
+                            logger.error(f"Error parsing date strings: {e}")
+                            return (3600000, None)
+                    else:
+                        # Assume numeric timestamps
+                        from_ts = int(from_val)
+                        to_ts = int(to_val)
+                        window_size_ms = to_ts - from_ts
+                        logger.info(f"Parsed timestamp range: from={from_ts}, to={to_ts}, window={window_size_ms}ms")
+                        return (window_size_ms, to_ts)
+                else:
+                    logger.warning("Absolute time range missing 'from' or 'to', defaulting to 1h")
+                    return (3600000, None)
+
+            # Handle string format for relative time range
+            time_str = str(time_input).lower().strip()
+            if time_str.endswith('h'):
+                hours = int(time_str[:-1])
+                window_size_ms = hours * 3600000
+            elif time_str.endswith('m'):
+                minutes = int(time_str[:-1])
+                window_size_ms = minutes * 60000
+            elif time_str.endswith('d'):
+                days = int(time_str[:-1])
+                window_size_ms = days * 86400000
+            else:
+                # Default to 1 hour if format not recognized
+                window_size_ms = 3600000
+
+            logger.info(f"Parsed relative time range '{time_str}' to {window_size_ms}ms")
+            return (window_size_ms, None)
+
+        window_size_ms, to_timestamp = parse_time_range(time_range)
+
+        # Build tag filter expression
+        if filters and len(filters) > 0:
+            # Single filter case - use TagFilter directly
+            if len(filters) == 1:
+                filter_dict = filters[0]
+                filter_name = filter_dict.get("name")
+                filter_value = filter_dict.get("value")
+
+                tag_filter_expression = TagFilter(
+                    type="TAG_FILTER",
+                    entity="NOT_APPLICABLE",
+                    name=filter_name,
+                    operator="EQUALS",
+                    value=TagFilterAllOfValue(filter_value)
+                )
+            else:
+                # Multiple filters - use TagFilterExpression with AND
+                tag_filter_elements = []
+                for filter_dict in filters:
+                    filter_name = filter_dict.get("name")
+                    filter_value = filter_dict.get("value")
+                    if filter_name and filter_value:
+                        tag_filter_elements.append(TagFilter(
+                            type="TAG_FILTER",
+                            entity="NOT_APPLICABLE",
+                            name=filter_name,
+                            operator="EQUALS",
+                            value=TagFilterAllOfValue(filter_value)
+                        ))
+
+                tag_filter_expression = TagFilterExpression(
+                    type="EXPRESSION",
+                    logical_operator="AND",
+                    elements=tag_filter_elements
+                )
+        else:
+            # No filters - use empty expression
+            tag_filter_expression = TagFilterExpression(
+                type="EXPRESSION",
+                logical_operator="AND",
+                elements=[]
+            )
+
+        # Build metrics array with proper Pydantic objects
+        # Note: granularity is intentionally omitted as it causes data discrepancies with UI
+        infra_metrics = []
+        for metric_name in metrics:
+            metric_config = SimpleMetricConfiguration(
+                metric=metric_name,
+                aggregation=aggregation.upper()
+            )
+            infra_metric = InfraMetricConfiguration(actual_instance=metric_config)
+            infra_metrics.append(infra_metric)
+
+        logger.info(f"Built {len(infra_metrics)} metric configurations (without granularity)")
+
+        # Build Order object if provided
+        order_obj = None
+        if order and isinstance(order, dict):
+            order_by = order.get("by")
+            order_direction = order.get("direction", "DESC").upper()
+            if order_by:
+                # Ensure order.by is in format "metricName.AGGREGATION"
+                # If LLM provides just metric name, append aggregation
+                if "." not in order_by and aggregation:
+                    order_by = f"{order_by}.{aggregation.upper()}"
+                    logger.info(f"Appended aggregation to order.by: {order_by}")
+
+                order_obj = Order(by=order_by, direction=order_direction)
+                logger.info(f"Built Order: by={order_by}, direction={order_direction}")
+
+        # Build pagination object
+        # Supports pagination formats: {"page": 1, "pageSize": 20} or {"offset": 0, "limit": 20}
+        page_size = 50  # Default
+        offset = None
+
+        if pagination and isinstance(pagination, dict):
+            # Handle pageSize or limit
+            page_size = pagination.get("pageSize") or pagination.get("limit", 50)
+
+            # Handle page number (convert to offset)
+            page = pagination.get("page")
+            if page is not None and page > 0:
+                offset = (page - 1) * page_size
+                logger.info(f"Converted page {page} to offset {offset}")
+
+            # Handle direct offset
+            if "offset" in pagination:
+                offset = pagination.get("offset", 0)
+
+            logger.info(f"Pagination: pageSize={page_size}, offset={offset}")
+
+        # Build CursorPagination object
+        if offset is not None and offset > 0:
+            cursor_pagination = CursorPagination(retrieval_size=page_size, offset=offset)
+        else:
+            cursor_pagination = CursorPagination(retrieval_size=page_size)
+
+        # Build TimeFrame object - supports both relative and absolute time ranges
+        if to_timestamp is not None:
+            # Absolute time range: from specific timestamp to another timestamp
+            time_frame = TimeFrame(window_size=window_size_ms, to=to_timestamp)
+            logger.info(f"Built absolute TimeFrame: window_size={window_size_ms}ms, to={to_timestamp}")
+        else:
+            # Relative time range: last N hours/minutes/days from now
+            time_frame = TimeFrame(window_size=window_size_ms)
+            logger.info(f"Built relative TimeFrame: window_size={window_size_ms}ms")
+
+        # Build query with proper Pydantic objects - conditional based on groupBy
+        if group_by and len(group_by) > 0:
+            # Use GetInfrastructureGroupsQuery for grouped queries
+            query = GetInfrastructureGroupsQuery(
+                group_by=group_by,  # Required for groups API
+                type=entity_type,
+                metrics=infra_metrics,
+                tag_filter_expression=tag_filter_expression,  # Required field
+                time_frame=time_frame,
+                pagination=cursor_pagination,
+                order=order_obj  # Optional
+            )
+            logger.info(f"Built GROUPS query payload for {entity_type} with groupBy={group_by}")
+        else:
+            # Use GetInfrastructureQuery for non-grouped queries (existing behavior)
+            query = GetInfrastructureQuery(
+                type=entity_type,
+                metrics=infra_metrics,
+                tag_filter_expression=tag_filter_expression,  # Required field
+                time_frame=time_frame,
+                pagination=cursor_pagination,
+                order=order_obj  # Optional
+            )
+            logger.info(f"Built ENTITIES query payload for {entity_type}")
+
+        # Log the complete payload being sent to API
+        import json
+        try:
+            payload_dict = query.to_dict()
+            logger.info("=" * 80)
+            logger.info("FINAL API PAYLOAD:")
+            logger.info(json.dumps(payload_dict, indent=2))
+            logger.info("=" * 80)
+        except Exception as e:
+            logger.warning(f"Could not serialize payload for logging: {e}")
+
+        # Call Instana API - conditional based on groupBy
+        try:
+            if group_by and len(group_by) > 0:
+                logger.info("Calling Instana API: get_entity_groups_without_preload_content (with groupBy)")
+                raw_response = api_client.get_entity_groups_without_preload_content(
+                    get_infrastructure_groups_query=query
+                )
+            else:
+                logger.info("Calling Instana API: get_entities_without_preload_content (no groupBy)")
+                raw_response = api_client.get_entities_without_preload_content(
+                    get_infrastructure_query=query
+                )
+            logger.info("API call successful, processing response...")
+
+            # Parse response
+            import json
+            response_text = raw_response.data.decode('utf-8')
+            result_dict = json.loads(response_text)
+
+            items = result_dict.get("items", [])
+            logger.info(f"API returned {len(items)} items")
+
+            # Format results - different format for grouped vs individual entities
+            if not items:
+                result_text = f"No {entity_type} entities found matching your criteria."
+            elif group_by and len(group_by) > 0:
+                # Grouped results format
+                result_text = f"Found {len(items)} groups (grouped by {', '.join(group_by)}):\n\n"
+                for idx, item in enumerate(items, 1):  # Show all groups
+                    # For groups, extract the group key from 'tags' field
+                    tags = item.get("tags", {})
+                    count = item.get("count", 0)
+                    metrics_data = item.get("metrics", {})
+
+                    # Build group label from tags (the groupBy fields)
+                    group_parts = []
+                    for tag_name in group_by:
+                        tag_value = tags.get(tag_name, "unknown")
+                        group_parts.append(f"{tag_name}={tag_value}")
+                    group_label = ", ".join(group_parts)
+
+                    result_text += f"{idx}. Group: {group_label} (count: {count})\n"
+                    if metrics_data:
+                        for metric_key, metric_value in metrics_data.items():
+                            # Check if metric_key matches any of the requested metrics
+                            if any(requested_metric in metric_key for requested_metric in metrics):
+                                result_text += f"   {metric_key}: {metric_value}\n"
+                    result_text += "\n"
+            else:
+                # Individual entities format - return all entities
+                result_text = f"Found {len(items)} {entity_type} entities:\n\n"
+                for idx, item in enumerate(items, 1):  # Show all entities
+                    label = item.get("label", "unknown")
+                    metrics_data = item.get("metrics", {})
+
+                    result_text += f"{idx}. {label}\n"
+                    if metrics_data:
+                        for metric_key, metric_value in metrics_data.items():
+                            # Check if metric_key matches any of the requested metrics
+                            if any(requested_metric in metric_key for requested_metric in metrics):
+                                result_text += f"   {metric_key}: {metric_value}\n"
+                    result_text += "\n"
+
+            return [TextContent(type="text", text=result_text)]
+
+        except Exception as e:
+            logger.error(f"API call failed: {e}", exc_info=True)
+            return [TextContent(
+                type="text",
+                text=f"Error calling Instana API: {e!s}"
+            )]

src/infrastructure/{infrastructure_analyze.py → infrastructure_analyze_old.py}

@@ -52,10 +52,7 @@ class InfrastructureAnalyzeMCPTools(BaseInstanaClient):
         """Initialize the Infrastructure Analyze MCP tools client."""
         super().__init__(read_token=read_token, base_url=base_url)
 
-    @register_as_tool(
-        title="Get Available Metrics",
-        annotations=ToolAnnotations(readOnlyHint=True, destructiveHint=False)
-    )
+    # @register_as_tool(...)  # Disabled for future reference
     @with_header_auth(InfrastructureAnalyzeApi)
     async def get_available_metrics(self,
                                     payload: Optional[Union[Dict[str, Any], str]] = None,
@@ -191,10 +188,6 @@ class InfrastructureAnalyzeMCPTools(BaseInstanaClient):
             logger.error(f"Error in get_available_metrics: {e}", exc_info=True)
             return {"error": f"Failed to get available metrics: {e!s}"}
 
-    @register_as_tool(
-        title="Get Entities",
-        annotations=ToolAnnotations(readOnlyHint=True, destructiveHint=False)
-    )
     @with_header_auth(InfrastructureAnalyzeApi)
     async def get_entities(self,
                            payload: Optional[Union[Dict[str, Any], str]] = None,
@@ -309,10 +302,6 @@ class InfrastructureAnalyzeMCPTools(BaseInstanaClient):
             logger.error(f"Error in get_entities: {e}", exc_info=True)
             return {"error": f"Failed to get entities: {e!s}"}
 
-    @register_as_tool(
-        title="Get Aggregated Entity Groups",
-        annotations=ToolAnnotations(readOnlyHint=True, destructiveHint=False)
-    )
     @with_header_auth(InfrastructureAnalyzeApi)
     async def get_aggregated_entity_groups(self,
                                            payload: Optional[Union[Dict[str, Any], str]] = None,
@@ -513,10 +502,6 @@ class InfrastructureAnalyzeMCPTools(BaseInstanaClient):
                 "error": f"Failed to summarize results: {e!s}"
             }
 
-    @register_as_tool(
-        title="Get Available Plugins",
-        annotations=ToolAnnotations(readOnlyHint=True, destructiveHint=False)
-    )
     @with_header_auth(InfrastructureAnalyzeApi)
     async def get_available_plugins(self,
                                     payload: Optional[Union[Dict[str, Any], str]] = None,