Crossplane ArgoCD Provider
You set up the crossplane argocd provider by installing its package into your Kubernetes cluster. Then you link it to your deployment server using a stored secret key. Platform teams use this setup to manage apps, projects, and cluster links from one control plane. Still, teams often grant the controller broad admin roles to make setup fast. When a shared role has too much power, one bad manifest can break live apps. How the crossplane argocd provider works in a cluster This extension lets yo
Your coding agents run with your access.
Claude Code reaches prod with a developer's credentials, and approval prompts get clicked through. hoop.dev checks every query, masks sensitive data, and holds risky changes for a human.
You set up the crossplane argocd provider by installing its package into your Kubernetes cluster. Then you link it to your deployment server using a stored secret key. Platform teams use this setup to manage apps, projects, and cluster links from one control plane. Still, teams often grant the controller broad admin roles to make setup fast. When a shared role has too much power, one bad manifest can break live apps.
How the crossplane argocd provider works in a cluster
This extension lets your control plane talk to your GitOps server API. Instead of clicking in a web UI, your team writes declarative resource files. The controller reads these files and calls the server API to make changes. It watches the state of your projects, repos, and apps all day. If someone changes a setting by hand, the controller fixes it right away.
To make this work, you first install the package in your main cluster. Next, you create an account token inside your deployment server. You save that token as a Kubernetes secret inside your control cluster. Then you point the provider config to read that secret key. Once linked, the controller can add target clusters, register Git repos, and sync apps.
Managing permissions and secrets across clusters
Every resource the controller makes needs the right level of access. Most teams start by giving the controller a single admin key. That key lets the control plane build projects and link new clusters fast. It also lets the controller read and write repo keys across every project. While this single key is easy to set up, it creates a big risk.
Anyone who can write a custom resource in the cluster uses that same admin key. A developer could change a project rule or point an app to the wrong cluster. To reduce this risk, teams split access by using scoped server roles. They give the controller a smaller role that only manages specific projects. They also use separate namespaces to keep team manifests apart from each other.
Crossplane Composition Security + ArgoCD Security: architecture patterns and best practices
Continue reading? Get the full guide.
Free. No spam. Unsubscribe anytime.
What native setup solves and where gaps remain
This native workflow solves the problem of manual setup for new clusters and apps. When you spin up a new cluster, the control plane registers it right away. It also sets up the base projects and links the right Git repos. Your team gets a repeatable way to bootstrap whole platforms from code. Cluster state stays in sync with your desired rules at all times.
Still, native role settings only decide who can log in and start a sync. They do not inspect what individual commands or database calls do in flight. And when an engineer needs to fix a broken app, they still need direct cluster access. Handing out static kubeconfig files or shared database keys creates new audit gaps. Native sync tools cannot mask sensitive fields like emails or card numbers during live debugging.
Govern kubectl and API calls in flight
Identity providers and Kubernetes RBAC decide who logs into a cluster. They cannot redact keys inside returned JSON objects or hold a risky delete call for review. By design, hoop.dev runs as an open-source proxy, a sidecar for your Kubernetes API, checking requests and responses without replacing RBAC.
First, hoop.dev guardrails refuse banned methods, paths, or headers before they run. Next, hoop.dev data masking rewrites sensitive keys in returned Secret, ConfigMap, and log responses in memory, and writes each decision to the audit trail. Explore the repo on GitHub.
Frequently asked questions
How does the controller log in to the deployment server?
The controller reads a secret stored inside your Kubernetes cluster. This secret holds an API token or login key from the deployment server. The controller sends this token with every API call to prove its identity.
Can this setup manage more than one deployment server?
Yes, you can create multiple config objects inside the same control cluster. Each config object points to a different server address and uses its own secret key. When you write a resource file, you specify which config object it should use.
What happens if someone changes an app setting in the web UI?
The controller checks the live state against your resource files on a regular loop. If it finds a manual change in the UI, it overwrites that change. The settings in your cluster manifest always win over manual edits.
Related guides
In this post