Skip to content
Design partners

ICE design partners

Run the loop on your cluster before the public release.

or write to [email protected]

Design partners ​

ICE is built on real estates, not fixtures. A design partner runs the loop on their own Kubernetes before the public release, and tells me where it is noisy, missing or wrong.

What you get ​

  • The loop on your estate early — findings and drift turned into pull requests you review, and cases that close only when a fresh measurement proves the fix.
  • A direct line to the engineer building it. Your feedback goes straight into the backlog.
  • A say in what gets built next. Capabilities move from tested to live on clusters like yours — that is exactly what the pilot is for.

How a pilot goes ​

  1. Try it locallykind + a synthetic company profile, 15–20 minutes
  2. Read-only agentin a test or staging cluster; it pushes facts out
  3. Review findingspersonin the console, with the reasons behind each
  4. Approve a changepersona rendered, digest-pinned pull request
  5. Merge and measuremeasurementthe case closes only on a fresh measurement

What leaves your cluster, and what can change it ​

  • The agent runs inside your cluster with list on workloads, network policies, config maps, mesh policies and policy reports — no secrets, no exec, no port-forward, no writes — and sends facts out. ICE never holds a kubeconfig for your clusters.
  • The operator can write only to ICE's own resources. Every change to your workloads is a pull request to your GitOps repository that you merge and your pipeline applies. You approve each one — the only exception is a trap response, and only if you opt in.
  • To confirm a merge was deployed, the operator reads Argo CD Application status (get and list, nothing else).
  • The optional model has no tools, PII redaction cannot be turned off, and the provider address has to be allow-listed by whoever runs the installation. The MCP interface is read-only and off by default.
  • One installation is one trust domain; there is no multi-tenancy yet.

What it needs ​

You runNeeded?
Kubernetes (managed or bare-metal)yes
A GitOps repository on GitHubyes — GitLab is tested, not yet live
Argo CDto let ICE confirm a merge was deployed (tested, not yet live)
Image scans in CI (Trivy or Grype)for vulnerability findings
Falco or Tetragonfor runtime signals
Prometheus and Alertmanager, Kyverno, Istio or Linkerdoptional — each adds detectors or posture facts

Interested? ​

Tell me about your estate — the cluster count, the GitOps setup, and what you would want ICE to catch first.

Updated:

ICE and deusex are in active development · KubeMeridian is archived (Apache-2.0) · RSS