The Control Plane for AI Infrastructure

Kubernetes-native registry for MCP servers, agents, skills & models

🚀 Quick Start • ✨ Features • 🏗️ Architecture • 📚 Docs • 📝 Blog

✨ What You Get

Automatically indexes MCP servers, agents, skills, and models across clusters.
If it's running, it's in the catalog.

No CLI needed. Agent Registry ships an MCP server — connect it to Claude Code, Cursor, or any MCP-compatible client and manage your registry conversationally.

Capability	What It Means
🔍 Auto-Discovery	Scans your clusters for AI workloads — MCP servers, agents, skills, models — and catalogs them automatically. Zero manual work.
📦 Unified Inventory	Everything in one place across dev, staging, prod. Git as the single source of truth.
✍️ Create & Publish	Generate manifests via UI/API, submit for review, open PRs — or deploy directly.
🚀 One-Click Deploy	Deploy from catalog to any environment. Controller handles the lifecycle.
🔒 GitOps Native	GitOps and Gitless Ops workflows built-in.
🌐 Multi-Cluster	Discover and deploy across clusters with workload identity.

🚀 Quick Start

One command to run dev environment

hljs language-bash

git clone https://github.com/den-vasyliev/agentregistry-inventory.git
cd agentregistry-inventory && make dev

🎯 That's it. UI opens at http://localhost:3000 with sample data pre-loaded.

No Kubernetes cluster needed — uses envtest (embedded etcd + kube-apiserver).

☸️ Have a cluster?

hljs language-bash

kubectl apply -k https://github.com/den-vasyliev/agentregistry-inventory/config/crd
helm install agentregistry-inventory ./charts/agentregistry -n agentregistry --create-namespace

🌟 Why Agent Inventory?

Without Agent Inventory	With Agent Inventory
😵 Sprawl of AI tools across clusters	📦 Single source of truth
🔍 Manual discovery of MCP servers	🤖 Auto-discovery & cataloging
😰 No version control for AI configs	📝 GitOps-native workflows
🤷 "What agents are running in prod?"	📊 Real-time inventory & status
😱 Direct K8s yaml edits	🚀 One-click deploy from UI/API

🏗️ Architecture

hljs language-vbnet

┌─────────────────────────────────────────────────────────────┐
│                      WEB UI (Next.js)                       │
│              embedded in controller at :8080                │
└───────────────────────────┬─────────────────────────────────┘
                            │ REST
                            ▼
┌─────────────────────────────────────────────────────────────┐
│                    CONTROLLER (Go Controller Runtime)       │
│  ┌─────────────┐ ┌────────────────┐ ┌─────────────────────┐ │
│  │  HTTP API   │ │  Reconcilers   │ │  Auto-Discovery     │ │
│  │   :8080     │ │                │ │                     │ │
│  └─────────────┘ └────────────────┘ └─────────────────────┘ │
│  ┌─────────────┐ ┌─────────────┐ ┌─────────────────────┐    │
│  │  MCP Server │ │   Metrics   │ │   Health            │    │
│  │   :8083     │ │   :8081     │ │   :8082             │    │
│  └─────────────┘ └─────────────┘ └─────────────────────┘    │
└───────────────────────────┬─────────────────────────────────┘
                            │ K8s API
                            ▼
┌─────────────────────────────────────────────────────────────┐
│                     CRDs (etcd)                             │
│  ┌─────────────────┐ ┌─────────────────┐ ┌──────────────┐   │
│  │ MCPServerCatalog│ │  AgentCatalog   │ │ SkillCatalog │   │
│  └─────────────────┘ └─────────────────┘ └──────────────┘   │
│  ┌────────────────────┐ ┌─────────────────┐                 │
│  │RegistryDeployment  | │ DiscoveryConfig │                 │
│  └────────────────────┘ └─────────────────┘                 │
└─────────────────────────────────────────────────────────────┘
                            │
                            ▼
┌─────────────────────────────────────────────────────────────┐
│              RUNTIME (Kagent + KMCP +...)                   │
│         Agent ↔ MCP Server ↔ Model ↔ Skills                 │
└─────────────────────────────────────────────────────────────┘

📚 CRD Reference

Kgateway — Gateway API for AI traffic
Kagent — Kubernetes AI agent runtime
KMCP — MCP server operator

📋 Inventory CRD Examples

📦 Publish a Catalog Entry

hljs language-yaml

apiVersion: agentregistry.dev/v1alpha1
kind: MCPServerCatalog          # Also: AgentCatalog, SkillCatalog, ModelCatalog
metadata:
  name: filesystem-v1-0-0
  namespace: agentregistry
spec:
  name: "filesystem"
  version: "1.0.0"
  title: "Filesystem MCP Server"
  description: "Provides file system access tools"
  websiteUrl: "https://github.com/modelcontextprotocol/servers"
  repository:
    url: "https://github.com/modelcontextprotocol/servers"
    source: github
  packages:
    - registryType: npm
      identifier: "@modelcontextprotocol/server-filesystem"
      version: "0.6.1"
      transport:
        type: stdio
  remotes:                        # Optional: streamable-http endpoints
    - type: streamable-http
      url: "https://mcp.example.com/filesystem"

🚀 Deploy to Runtime

hljs language-yaml

apiVersion: agentregistry.dev/v1alpha1
kind: RegistryDeployment
metadata:
  name: filesystem-deployment
  namespace: agentregistry
spec:
  resourceName: "filesystem"
  version: "1.0.0"
  resourceType: mcp             # mcp | agent
  runtime: kubernetes           # Required: deployment runtime
  namespace: default            # Target namespace
  preferRemote: false           # Use local package vs remote endpoint
  environment: ""               # Target environment (from DiscoveryConfig), empty = local cluster
  config:                       # Optional: deployment configuration
    LOG_LEVEL: "info"

The controller reconciles this → creates MCPServer/Agent CRs → tracks status.

🌍 Multi-Cluster Discovery

hljs language-yaml

apiVersion: agentregistry.dev/v1alpha1
kind: DiscoveryConfig
metadata:
  name: multi-cluster-discovery
  namespace: agentregistry
spec:
  environments:
    - name: production
      cluster:
        name: prod-gke
        projectId: my-gcp-project
        zone: us-central1
        useWorkloadIdentity: true
      provider: gcp
      discoveryEnabled: true
      deployEnabled: false
      namespaces: [ai-workloads, agents]
      resourceTypes: [MCPServer, Agent, ModelConfig]

→ Full Autodiscovery Docs

🔌 API Reference

Public API (Read-Only, no auth required)

hljs language-bash

curl http://localhost:8080/v0/servers
curl http://localhost:8080/v0/agents
curl http://localhost:8080/v0/skills

Admin API (Write)

hljs language-bash

# Auth disabled by default — just POST
curl -X POST http://localhost:8080/admin/v0/servers \
  -H "Content-Type: application/json" \
  -d @server.json

# With auth enabled (AGENTREGISTRY_AUTH_ENABLED=true)
curl -X POST http://localhost:8080/admin/v0/servers \
  -H "Authorization: Bearer your-token" \
  -H "Content-Type: application/json" \
  -d @server.json

🤖 MCP Server

The controller embeds an MCP server on :8083. Connect any MCP-compatible client (Claude Code, Cursor, etc.) to browse, deploy, and manage registry resources conversationally.

Connect (auth disabled, default)

hljs language-json

{
  "mcpServers": {
    "agentregistry": {
      "type": "streamable-http",
      "url": "http://localhost:8083/mcp"
    }
  }
}

Connect (auth enabled)

hljs language-json

{
  "mcpServers": {
    "agentregistry": {
      "type": "streamable-http",
      "url": "http://localhost:8083/mcp",
      "headers": {
        "Authorization": "Bearer your-token"
      }
    }
  }
}

Tools

Tool	Description
`list_catalog`	List catalog entries (servers/agents/skills/models)
`get_catalog`	Get entry details
`get_registry_stats`	Counts of all resource types
`list_deployments`	List active deployments
`get_deployment`	Deployment details by name
`deploy_catalog_item`	Deploy a catalog item to Kubernetes
`delete_deployment`	Remove a deployment
`update_deployment_config`	Update deployment config
`list_environments`	Discovered environments from DiscoveryConfig
`get_discovery_map`	Cluster topology and resource counts
`trigger_discovery`	Force re-scan of discovery
`recommend_servers`	AI-powered server recommendations
`analyze_agent_dependencies`	AI-powered dependency analysis
`generate_deployment_plan`	AI-powered deployment planning

→ Full MCP Docs | → MCP Auth

☸️ Production Deployment

hljs language-bash

helm install agentregistry ./charts/agentregistry \
  --namespace agentregistry \
  --create-namespace \
  --set replicaCount=2 \
  --set controller.leaderElection=true

Key Configuration

Setting	Default	Description
`replicaCount`	`1`	Set to 2+ for HA
`controller.leaderElection`	`false`	Required for multi-replica
`controller.logLevel`	`info`	Use `debug` for troubleshooting
`httpApi.serviceType`	`ClusterIP`	Use `LoadBalancer` for external access
`disableAuth`	`true`	Set to `false` to enable Bearer token auth

🔐 Authentication

Auth is disabled by default. When enabled (disableAuth: false), the admin API and MCP server require a Bearer token from the agentregistry-api-tokens Kubernetes Secret.

For production deployments with a gateway, use agentgateway to handle Azure AD JWT validation in front of the controller — keep disableAuth: true on the backend and let the gateway enforce auth.

Enable Bearer Token Auth

hljs language-bash

# Create token secret
kubectl create secret generic agentregistry-api-tokens \
  -n agentregistry \
  --from-literal=admin-token=$(openssl rand -hex 32)

# Install with auth enabled
helm install agentregistry ./charts/agentregistry \
  --namespace agentregistry \
  --set disableAuth=false

Azure AD (MSAL Browser PKCE)

The embedded UI supports Azure AD login via MSAL.js — no client secret required.

hljs language-yaml

# charts/agentregistry/values.yaml
azure:
  tenantId: "your-tenant-id"
  clientId: "your-client-id"

→ Full Azure AD Setup

🧪 Development

hljs language-bash

make dev          # Full stack: controller + UI dev server
make run          # Controller only with embedded UI at :8080
make test         # Run test suite
make lint         # gofmt + go vet
make build        # Build UI + controller binary
make generate     # Regenerate CRD manifests + deepcopy

🔗 Ecosystem

Project	Description	Link
Agentgateway	MCP/HTTP auth gateway	agentgateway.dev
MCP	Model Context Protocol specification	modelcontextprotocol.io
Kagent	Kubernetes AI agent runtime	github.com/kagent-dev/kagent
KMCP	MCP server operator for Kubernetes	github.com/kagent-dev/kmcp
MCP Go SDK	Official Go SDK	github.com/modelcontextprotocol/go-sdk

agentregistry-inventory

The Control Plane for AI Infrastructure

Kubernetes-native registry for MCP servers, agents, skills & models

✨ What You Get

Automatically indexes MCP servers, agents, skills, and models across clusters.
If it's running, it's in the catalog.

🚀 Quick Start

One command to run dev environment

🌟 Why Agent Inventory?

🏗️ Architecture

📚 CRD Reference

📋 Inventory CRD Examples

📦 Publish a Catalog Entry

🚀 Deploy to Runtime

🌍 Multi-Cluster Discovery

🔌 API Reference

Public API (Read-Only, no auth required)

Admin API (Write)

🤖 MCP Server

Connect (auth disabled, default)

Connect (auth enabled)

Tools

☸️ Production Deployment

Key Configuration

🔐 Authentication

Enable Bearer Token Auth

Azure AD (MSAL Browser PKCE)

🧪 Development

🔗 Ecosystem

💬 Join the Community

Similar Packages

Original Author

Publisher

agentregistry-inventory

The Control Plane for AI Infrastructure

Kubernetes-native registry for MCP servers, agents, skills & models

✨ What You Get

Automatically indexes MCP servers, agents, skills, and models across clusters. If it's running, it's in the catalog.

🚀 Quick Start

One command to run dev environment

🌟 Why Agent Inventory?

🏗️ Architecture

📚 CRD Reference

📋 Inventory CRD Examples

📦 Publish a Catalog Entry

🚀 Deploy to Runtime

🌍 Multi-Cluster Discovery

🔌 API Reference

Public API (Read-Only, no auth required)

Admin API (Write)

🤖 MCP Server

Connect (auth disabled, default)

Connect (auth enabled)

Tools

☸️ Production Deployment

Key Configuration

🔐 Authentication

Enable Bearer Token Auth

Azure AD (MSAL Browser PKCE)

🧪 Development

🔗 Ecosystem

💬 Join the Community

Similar Packages

Original Author

Publisher

Automatically indexes MCP servers, agents, skills, and models across clusters.
If it's running, it's in the catalog.