Scale Argo CD across hundreds of clusters with a single pane of glass 🌐

Imagine managing GitOps deployments across edge locations, multiple cloud providers, air-gapped environments, and remote sites—all from one central dashboard. argocd-agent makes this reality by extending Argo CD with a distributed architecture that brings the control plane to you, no matter where your workloads live.

The Challenge: Traditional multi-cluster Argo CD setups hit walls when scaling to hundreds of clusters, especially across unreliable networks, air-gapped environments, or edge locations.

The Solution: argocd-agent flips the script—instead of your control plane reaching out to remote clusters, lightweight agents reach back to a central hub. This "pull model" enables:

✅ Massive Scale: Manage thousands of applications across hundreds of clusters
✅ Network Resilience: Works with intermittent connections, high latency, or restricted networks
✅ Edge-Friendly: Perfect for IoT, retail, manufacturing, or remote deployments
✅ Air-Gap Ready: Secure deployments that never expose cluster internals
✅ Cloud Agnostic: Seamlessly span AWS, GCP, Azure, on-premises, and hybrid environments

• 🏭 Manufacturing: Deploy to factory floors and remote facilities
• 🛒 Retail: Manage point-of-sale and in-store systems across locations
• 🚢 Edge Computing: IoT deployments, autonomous vehicles, ships, and remote sites
• 🏛️ Enterprise: Multi-datacenter deployments with strict security requirements
• ☁️ Multi-Cloud: Unified GitOps across different cloud providers
• 🔒 Air-Gapped: Secure environments with restricted network access

Get up and running in minutes! Check out our Getting Started Guide for step-by-step instructions.

# 1. Set up the control plane (where your Argo CD UI lives)
kubectl apply -k https://github.com/argoproj-labs/argocd-agent/install/kubernetes/principal

# 2. Deploy agents to your workload clusters
kubectl apply -k https://github.com/argoproj-labs/argocd-agent/install/kubernetes/agent

# 3. Watch the magic happen ✨

Want to try it out? Our getting started guide gets you running with a local environment in under 10 minutes.

Think of argocd-agent as a hub-and-spoke architecture where agents reach back to the control plane:

    ┌─────────────────┐
    │  Control Plane  │  ← Your Argo CD UI and API
    │   (The Hub)     │     (No outbound connections needed!)
    └─────────────────┘
              ▲ ▲ ▲
              │ │ │
              │ │ │
┌─────────────┘ │ └─────────────┐
│               │               │
│               │               │
┌─────────┐ ┌─────────┐ ┌─────────┐
│ Agent 1 │ │ Agent 2 │ │ Agent N │  ← Each agent connects independently
│ AWS     │ │ Factory │ │ Edge    │     (Pull model - no inter-agent links!)
└─────────┘ └─────────┘ └─────────┘

🎛️ Control Plane: Your familiar Argo CD interface—manage everything from one place
🤖 Agents: Lightweight components that reach out and connect to the hub
🔄 Smart Sync: Agents pull configuration and push status updates automatically

🎯 Managed Mode: Perfect for centralized control

• Deploy applications from your control plane to remote clusters
• Ideal for rolling out updates, managing configurations, and maintaining consistency

🦾 Autonomous Mode: Built for independence

• Remote clusters manage their own applications (via GitOps)
• Control plane provides observability and monitoring
• Perfect for air-gapped or highly autonomous environments

Mix and match modes across your fleet—some clusters managed, others autonomous, all visible from one dashboard.

• mTLS everywhere: All communications are encrypted and authenticated
• Zero trust: Control plane never needs direct cluster access
• Certificate-based auth: Strong identity verification for every agent

• Intermittent connections: Agents work offline and sync when possible
• High latency tolerant: Designed for satellite links, cellular, and unreliable networks
• HTTP/1.1 compatible: Works through corporate proxies and legacy infrastructure

• Single pane of glass: See all clusters, applications, and deployments in one view
• Real-time status: Health, sync status, and metrics from all environments
• Live resources: Inspect Kubernetes resources across your entire fleet

• Lightweight: Minimal resource footprint on remote clusters
• Self-healing: Agents automatically reconnect and recover
• Easy upgrades: Rolling updates without downtime

🌱 Early Days: argocd-agent is in active development and not yet ready for production use. We're building in the open to collaborate with the community and gather feedback early.

What works today:

• ✅ Basic hub-and-spoke architecture
• ✅ Application deployment and status sync
• ✅ Multi-cluster observability
• ✅ Both managed and autonomous modes
• ✅ Live resource viewing and manipulation
• ✅ Custom resource actions
• ✅ Resource proxy for transparent cluster access

Coming soon:

• 🚧 Pod logs streaming
• 🚧 Advanced RBAC and multi-tenancy
• 🚧 Enhanced UI integrations
• 🚧 Production hardening

Track our progress in the roadmap and milestones.

We're building argocd-agent together! Whether you're a GitOps veteran or just getting started, there are many ways to contribute:

💬 Get Help & Share Ideas

• GitHub Discussions - Ask questions, share use cases
• #argo-cd-agent on CNCF Slack - Real-time chat

🛠️ Contribute

• Contributing Guide - Code, docs, and testing guidelines
• Issue Tracker - Bug reports and feature requests
• Good First Issues - Perfect for newcomers

📖 Learn More

• Documentation - Comprehensive guides and references
• Architecture Deep Dive - Understanding the internals
• Configuration Guide - Detailed setup instructions

argocd-agent is not yet production-ready but we're working hard to get there! If you're interested in:

• 🧪 Testing in dev/staging environments
• 🤝 Contributing to development
• 💼 Enterprise partnerships
• 🗣️ Speaking about your use case

We'd love to hear from you! Start a discussion or reach out on Slack.

argocd-agent is licensed under the Apache License 2.0.