---

name: mcp-tools-reference description: MCP tools quick reference for Giant Swarm operations

MCP Tools Quick Reference

Discovery

• mcp_muster_list_core_tools: List core platform tools
• mcp_muster_filter_tools: Search tools by pattern (e.g., workflow_*, x_kubernetes*, *capi*, *_teleport_*)
• core_workflow_list: List available workflows

Execution

All tools execute via: mcp_muster_call_tool

• core_* tools: Core platform building blocks
• x_* tools: Extended tools (kubernetes, prometheus, teleport, flux, capi)
• workflow_* tools: High-level automated workflows (composable)

Critical Parameters

Kubernetes (x_kubernetes_*):

• kubeContext: REQUIRED - Always set explicitly (current context ignored)
• Format: teleport.giantswarm.io-{mc} or teleport.giantswarm.io-{mc}-{wc}

Prometheus (x_prometheus_*):

• prometheus_url: REQUIRED - Format: http://localhost:{localPort}/prometheus
• Port from workflow_login-management-cluster (unique per MC)
• Verify port works before queries

Teleport (x_teleport_*, workflows):

• localPort: MUST be string, not integer (e.g., "18001", not 18001)

CAPI (*capi*):

• MC context only - Tools fail on WC context

Common Workflows

Login Management Cluster:

• Tool: workflow_login-management-cluster
• Args: installation (MC name), localPort (string)
• Creates: kubectl context + Mimir port-forward

Login Workload Cluster:

• Tool: workflow_login-workload-cluster
• Args: installation (MC name), localPort (string), workloadCluster (WC name)
• Creates: kubectl context

List Clusters:

• Tool: x_teleport_kube_list_clusters
• Returns: All available MC/WC clusters

Tool Categories

• workflow_*: Automated multi-step operations (prefer these)
• x_kubernetes_*: K8s resource operations, logs, exec
• x_prometheus_*: PromQL queries, metric fetching
• x_teleport_*: Cluster discovery, access management
• x_flux_*: GitOps operations, Kustomization inspection
• *capi*: Cluster API operations (MC only)
• core_*: Platform primitives (workflow building blocks)

Workflow Debugging

• Inspect: core_workflow_get
• Execution history: core_workflow_execution_list
• On error: Check workflow definition + execution logs

Best Practices

1. Check existing workflows first - Use mcp_muster_filter_tools with workflow_* pattern
2. Prefer MCP tools over kubectl/CLI - Better integration, error handling
3. Create workflows from multi-step tasks - Makes platform composable
4. Verify ports before Prometheus queries - Login workflow creates port-forward
5. Always set kubeContext - Never rely on current context