Ecosystem & Marketplaces›Open-Source MCP Servers (10)›ArgoCD MCP Server›Resources & prompts

Resources and prompts

Alongside its 29 tools, the ArgoCD MCP Server exposes 5 read-only argocd:// resources for real-time state observation and 7 guided MCP prompts that encode standard GitOps operational runbooks.

MCP resources (argocd://)

Resources provide side-effect-free snapshots of application health, active sync operations, and deployment events.

Resource URI	Refresh Cadence	Description
argocd://applications/{cluster}	5s	Inventory of all applications on the target cluster with health, sync status, and target revision
argocd://application-metrics/{cluster}/{app}	10s	Pod readiness, replica counts, resource utilization, and sync duration for a specific application
argocd://sync-operations/{cluster}	2s	Active in-flight sync operations, sync phases, and resource apply progress
argocd://deployment-events/{cluster}	Real-time	Stream of ArgoCD and Kubernetes deployment events across the cluster
argocd://cluster-health/{cluster}	30s	Aggregate counts of Healthy, Degraded, Progressing, and OutOfSync applications

argocd://applications/{cluster}

Returns all applications targeting {cluster} (for example, argocd://applications/production), including each application's health.status, sync.status, last sync timestamp, and Git revision.

argocd://application-metrics/{cluster}/{app}

Returns pod health breakdowns, replica counts, and sync duration metrics for a single application (for example, argocd://application-metrics/production/payment-service).

argocd://sync-operations/{cluster}

Streams active sync operations so an agent can monitor a multi-resource rollout until all health checks report Healthy.

Guided MCP prompts (7)

MCP prompts provide structured multi-tool playbooks that guide the AI agent through complex GitOps operations in a deterministic sequence:

Prompt Name	Tools Orchestrated	Operational Purpose
onboard_github_repository	4 tools	Verify environment credentials, test connectivity, register a GitHub repo, and confirm registration
full_application_deployment	11 tools	End-to-end workflow: onboard repo, create app, preview manifest diff, sync, monitor, and verify health
debug_application_issues	5 tools	Inspect degraded resources, extract pod error logs, correlate Kubernetes events, and recommend fixes
rollback_decision	8 tools	Inspect revision history, preview rollback diff, disable auto-sync if needed, roll back, and verify recovery
setup_argocd_project	4 tools	Configure an AppProject with scoped source repos, destination namespaces, and RBAC policies
deploy_new_version	7 tools	Validate target revision, preview manifest diff, execute sync, and monitor rollout completion
post_deployment_validation	4 tools	Verify post-sync health status, resource readiness, and absence of pod error logs

Example: debug_application_issues execution flow

When you prompt an agent to debug a failing application ("My app payment-service is Degraded in production, help me debug it"), the debug_application_issues prompt guides the agent through five deterministic steps:

1. Call get_application_details to isolate which child resources (Deployment, Pod, Service) are unhealthy.
2. Call get_application_logs to extract error stack traces and connection failures from failing pods.
3. Call get_application_events to check for CrashLoopBackOff, ImagePullBackOff, or FailedMount events.
4. Synthesize the root cause across manifest state, pod logs, and cluster events.
5. Propose a concrete remediation diff or rollback plan.