Lumino MCP Server

Retrieve details about a Kubernetes/Tekton resource. Args: resource_type: Resource type. Supported: pod, service, configmap, secret, pvc, namespace, node, serviceaccount, endpoints, event, persistentvolume, resourcequota, limitrange, deployment, replicaset, daemonset, statefulset, job, cronjob, ingress, storageclass, hpa (horizontalpodautoscaler), pipelinerun, taskrun, pipeline, task, clustertask, triggertemplate, triggerbinding, eventlistener, podmonitor, servicemonitor, prom…

Monitor OpenShift Machine Config Pools for node configuration and update rollouts. Analyzes pool status, update progress, and configuration drift. Args: pool_names: Pools to monitor (default: all). include_node_details: Include node status per pool (default: True). include_update_history: Include update history (default: True). filter_updating: Only show updating pools (default: False). Returns: Dict: Keys: pools_overview, machine_config_pools, recent_config_changes, issues, update_recommendations.

Check health and status of OpenShift cluster operators for platform functionality. Analyzes operator conditions, versions, and dependencies. Args: operator_names: Operators to check (default: all). include_conditions: Include condition details (default: True). show_version_info: Include version info (default: True). filter_degraded: Only show operators with issues (default: False). include_dependencies: Show operator dependencies (default: False). Returns: Dict: Keys: cluster_info, operator_status, health_summary, critical_issues, dependencies.

Fetch logs from all pods in a Tekton PipelineRun with adaptive volume management. Prioritizes failed pods and manages token budgets automatically when no time/line filters specified. Args: pipelinerun_name: PipelineRun name. namespace: Kubernetes namespace. clean_logs: Clean and format logs (default: True). tail_lines: Lines from end (optional). since_seconds: Logs newer than N seconds (optional). since_time: Logs newer than RFC3339 timestamp (optional). timestamps: Include timestamps (default: True). previous: Get logs from previous container instance (defaul…

Get cluster-wide status summary of all Tekton PipelineRuns and TaskRuns. Shows running/succeeded/failed counts, recent failures, and long-running pipelines (>1 hour). Args: pipeline_runs_limit: Max PipelineRuns to fetch cluster-wide (default: 500). task_runs_limit_per_namespace: Max TaskRuns to fetch per namespace (default: 100). max_namespaces: Max namespaces to scan for TaskRuns (default: 20). recent_failures_limit: Max recent failures to include in output (default: 10). long_running_limit: Max long-running pipelines to include (default: 5). Returns: Dict[str, Any]…

Investigate TLS/certificate issues across the cluster with targeted search and analysis. Searches system namespaces for TLS error patterns and correlates with certificate events. Args: time_range: Search time range (default: "24h"). max_namespaces: Max namespaces to search (default: 20). focus_on_system_namespaces: Prioritize system namespaces (default: True). Returns: Dict: TLS issues, affected pods, certificate problems, and remediation suggestions.

List all namespaces in the Kubernetes cluster. Returns: List[str]: Alphabetically sorted namespace names. Empty list if access denied or cluster unreachable.

List Tekton PipelineRuns in a namespace with status and timing details. Args: namespace: Kubernetes namespace to query. limit: Maximum number of PipelineRuns to return (default: 200). Set to 0 for no limit. Returns: List[Dict]: PipelineRuns with keys: name, pipeline, status, started_at, completed_at, duration. Empty list if none found. [{"error": "msg"}] on failure.

List all pods in a Kubernetes namespace with status and placement info. Args: namespace: Kubernetes namespace to query. Returns: List[Dict]: Pods with keys: name, status, ip, node_name, creation_timestamp, restart_count, container_states (list of waiting/terminated reasons).

List recent Tekton PipelineRuns across all accessible namespaces, sorted by start time. Args: limit: Max PipelineRuns to retrieve (default: 10). Returns: Dict[str, List[Dict]]: Namespace to PipelineRun list. Each run has: namespace, name, start_time, status, pipeline, labels.

List Tekton TaskRuns in a namespace, optionally filtered by a specific PipelineRun. Args: namespace: Kubernetes namespace to query. pipeline_run: Optional PipelineRun name to filter by. Returns: List[Dict]: TaskRuns with keys: name, task, pipeline_run, status, started_at, completed_at, duration.

Generate real-time dependency graph of Kubernetes/Tekton components and their interconnections. Maps Services, Deployments, Pipelines, PVCs, and their relationships via ownerReferences and selectors. Args: cluster_names: Clusters to map (default: all). component_types: Filter by types (services, deployments, pipelines, pvcs, etc.). Note: secrets are NOT included by default. namespace_filter: Regex pattern to filter namespaces. depth_limit: Max dependency depth (default: 5). include_metrics: Include resource metrics (default: False). output_format: "json" (default), "g…

Manage training data for the predictive log analyzer. This tool allows viewing, collecting, and managing failure labels used for supervised learning in the predictive_log_analyzer. Args: action: Action to perform: - "stats": Get training data statistics (default) - "list_failures": List recent failure labels - "add_failure": Manually add a failure label - "collect": Trigger failure collection from namespaces - "cleanup": Remove old training data failure_type: For add_failure - type of failure (e.g., "crash", "oom", "image", "timeout") names…

Trace a logical operation (commit, PR, image) as it flows through pipelines. Correlates pipeline runs using labels, annotations, and artifact references. Args: trace_identifier: Commit SHA, PR number, image tag, or custom trace ID. trace_type: "commit", "pr", "image", or "custom". start_time: ISO 8601 start timestamp. end_time: ISO 8601 end timestamp. include_artifacts: Include artifact details (default: True). trace_depth: "shallow" or "deep" (default: "deep"). namespaces: Specific namespaces to search (skips auto-detection). max_namespaces: Maximum namespace…

Predict failures using ML analysis of historical log patterns before critical outages occur. Uses anomaly detection algorithms to correlate log patterns with failure events. Supports persistent model storage for faster subsequent calls. Args: prediction_window: Time window - "1h", "6h", "24h", "7d" (default: "6h"). confidence_threshold: Min confidence for predictions 0.0-1.0 (default: 0.75). log_sources: Sources to analyze - pods, services, nodes (default: all). namespaces: Specific namespaces to analyze (default: auto-detect active namespaces). max_namespaces: Maximum na…

Progressive event analysis with multiple detail levels and correlation detection. Args: namespace: Kubernetes namespace to analyze. analysis_level: "overview", "detailed", "correlation", or "deep_dive" (default: "overview"). time_period: Time window (e.g., "2h", "4h", "1d"). event_filters: Filters like {"severity": ["CRITICAL"], "category": ["FAILURE"]}. seed_event_id: Event ID for correlation analysis. focus_areas: Areas to emphasize (default: ["errors", "warnings", "failures"]). Returns: Dict: Analysis results based on selected level.

Execute PromQL queries against Prometheus for cluster metrics. Supports instant and range queries with automatic endpoint discovery and authentication. Args: query: PromQL query string. query_type: "instant" or "range" (default: "instant"). start_time: Start for range queries (ISO 8601 or Unix timestamp). end_time: End for range queries (ISO 8601 or Unix timestamp). step: Step interval for range queries (default: "300s"). cluster: Cluster domain override. format: "json", "table", or "csv" (default: "json"). namespace_filter: Regex to filter by namespace. l…

Query archived Kubernetes resources from KubeArchive (historical data no longer on the cluster). Single entry point for archived resources and their logs: set include_logs=True to attach logs for pipelinerun, taskrun, and pod results (use an exact name when you only need one resource). Optional container selects a container for multi-container pods. Args: resource_type: One of pipelinerun, taskrun, pod, release, snapshot (case-insensitive). namespace: Kubernetes namespace to search. name: Optional resource name; wildcards supported (e.g. my-pipeline-*). label_selector: Kubern…

Forecast resource bottlenecks by analyzing utilization trends and predicting exhaustion points. Uses time-series analysis to predict CPU, memory, disk, and network capacity constraints. Args: forecast_horizon: Forecast window - "1h", "6h", "24h", "7d", "30d" (default: "24h"). resource_types: Resources to analyze - cpu, memory, disk, network, pvc (default: all). namespaces: Specific namespaces to focus on. trend_analysis_period: Historical period for trends (default: "7d"). Returns: Dict: Keys: forecasts, capacity_recommendations, cluster_overview, historical_accuracy.

Search Kubernetes resources by labels across multiple resource types and namespaces. Args: resource_types: Types to search (e.g., ["pods", "services", "deployments"]). label_selectors: Criteria list [{"key": str, "value": str, "operator": "equals|exists|not_equals|in|not_in"}]. namespaces: Namespaces to search (default: all). limit_per_type: Max results per type (default: 100). include_metadata_only: Return only metadata (default: False). include_status: Include status info (default: True). sort_by: "name", "namespace", "creation_time", or "labels" (default: "creat…

Search logs using natural language queries with semantic understanding beyond keyword matching. Uses NLP for query interpretation, Kubernetes/Tekton entity recognition, and relevance ranking. Args: query: Natural language query describing what to search for. time_range: Time range - "1h", "6h", "24h", "7d" (default: "1h"). namespaces: Specific namespaces to search (default: auto-detect relevant namespaces). severity_levels: Log severity levels to include. max_results: Maximum results to return (default: 100). context_lines: Surrounding lines per match (default: 3). …

Adaptive event analysis for a namespace with automatic volume management. When no constraints specified, automatically: estimates volume, applies smart time windows, prioritizes errors/warnings, samples within token limits. Args: namespace: Kubernetes namespace to analyze. last_n_events: Exact event count (only if user specifies). time_period: Exact time window (only if user specifies). strategy: "auto" for adaptive behavior (default). focus_areas: Areas to emphasize (default: ["errors", "warnings", "failures"]). max_context_tokens: Max output tokens (default: 8000). …

Adaptive pod log analysis with automatic volume management and multi-pass processing. When no time constraints specified, automatically estimates volume and selects optimal time windows. Args: namespace: Kubernetes namespace. pod_name: Pod name to analyze. container_name: Specific container (if multiple). summary_level: "brief", "detailed", or "comprehensive" (default: "detailed"). focus_areas: Analysis focus (default: ["errors", "warnings", "performance"]). time_segments: Time-based segments to analyze (default: 5). max_context_tokens: Max tokens for analysis (de…

Stream and analyze pod logs in chunks with progressive pattern detection. Processes logs in manageable chunks for memory efficiency and real-time insights. Args: namespace: Kubernetes namespace. pod_name: Pod name to stream logs from. container_name: Specific container (if multiple). chunk_size: Lines per chunk (default: 5000). analysis_mode: "errors_only", "errors_and_warnings" (default), "full_analysis", or "custom_patterns". time_window: Time window for historical logs (e.g., "1h", "6h", "24h"). follow: Stream logs in real-time (default: False). max_chunks:…

Cluster pod logs into unique structural templates using Drain3 parse-tree mining. Unlike analyze_logs or analyze_pod_logs_hybrid which scan each line individually with regex, this tool groups thousands of repetitive log lines into a handful of distinct patterns (e.g. 10,000 lines → 25 templates). Use this when: - Logs are high-volume and noisy — the user needs to see the forest, not every tree - The user asks to "summarize", "deduplicate", "find patterns", or "reduce" logs - You need to know WHAT kinds of log lines exist and HOW MANY of each, not just errors Do NOT use this for quick error s…

Simulate impact of configuration changes before applying to live system with risk assessment. Uses Monte Carlo simulation and load modeling based on historical data. Args: scenario_type: Type - "resource_limits", "scaling", "configuration", "deployment". changes: Changes to simulate with before/after values. scope: Simulation scope - clusters, namespaces, components. simulation_duration: Duration - "1h", "24h", "7d" (default: "24h"). load_profile: Expected load - "current", "peak", "custom" (default: "current"). risk_tolerance: Risk level - "conservative", "moderate",…