Moncho
Moncho is a sovereign, managed-Kubernetes platform: production-grade clusters provisioned onto dedicated infrastructure through an open-source control plane, wrapped in a continuous compliance-evidence layer and a signed add-on supply chain built for banks, government, and regulated enterprises. It is delivered and operated end to end by Vigilus Labs.
Moncho is a sovereign managed-Kubernetes platform for banks, government, and regulated enterprises. It provisions clusters through an open-source control plane, then wraps them in a continuous compliance-evidence pack and a signed add-on supply chain — a cluster your regulator can accept.
What Moncho does
Kubernetes clusters are provisioned onto dedicated servers through an open-source control plane (k0rdent, k0smotron, Cluster API) and lifecycle-managed from one portal, so a tenant gets a running, governed cluster without operating the control plane themselves.
One organization login, with mandatory multi-factor authentication for financial-sector tenants, reaches every cluster the organization owns, and org admins self-service team-to-role-to-cluster access grants that become per-cluster RBAC, so cluster credentials are never handed out by hand.
A browser console inside the portal gives resource browsing, live log streaming, YAML apply, and an audited pod terminal by forwarding the tenant's own verified identity token to their cluster, so the tenant's own cluster does all the authorization and the console proxy carries no standing cluster credential.
A live evidence assembler continuously maps the platform's real security posture against major control frameworks (including the CIS Kubernetes Benchmark, ISO 27001, and Bangladesh regulatory control sets), tagging each control honestly as verified-live or attested rather than asserting blanket compliance.
Dated posture snapshots are sealed into a SHA-256 hash-chain and copied in real time into a write-once, object-locked store that stays unaltered even by an administrator, giving auditors a continuously-collectable, provably-unbackdated series.
A nightly job runs the CIS Kubernetes Benchmark (via kube-bench) against the platform and feeds the pass/fail result into the compliance pack as a measured control instead of an asserted one.
The compliance pack exports as an OSCAL 1.1.3 Assessment Results document so an auditor's own tooling can ingest the platform's control posture directly.
Every Day-2 add-on image is mirrored into a private registry, gated by two independent vulnerability scanners (Trivy and Grype) plus known-exploited-CVE and secret checks, and cosign-signed with a self-held key, so tenants never pull an unvetted chart off the public internet.
A Kyverno verifyImages policy runs inside each tenant cluster and rejects any catalog image that is not signed by the platform's own key, so signature enforcement happens at the tenant's own API server at deploy time.
Emergency access to a wedged tenant cluster is time-boxed, single-cluster, justification-required, and recorded on a tamper-evident chain, and privileged terminal sessions are recorded output-only (never keystrokes or typed secrets) and reviewable solely through a separated auditor role.
Three isolation tiers
Built on open source
Moncho is assembled entirely from open-source foundations, the k0rdent control-plane manager, k0smotron hosted control planes, and Cluster API, rather than a proprietary cloud stack. Its supply-chain trust, the signing and verification of add-on images, runs on self-held keys with no runtime dependency on external, public-good signing or transparency services. That is what lets the whole platform, and the clusters and data it manages, be operated as a self-contained sovereign estate by Vigilus Labs.
What does Moncho's compliance-evidence pack cover?
It maps each control to CIS Kubernetes Benchmark, ISO 27001, and Bangladesh-regulatory framework references, with continuous CIS benchmarking. It is a control crosswalk, not a certification — Moncho reports posture against these frameworks; it does not certify your cluster.
How isolated is each customer's cluster?
Isolation is a spectrum you choose. At the low end, a shared virtual cluster suits sandboxes and cost-sensitive workloads; at the high end, a whole cluster per tenant gets its own API server, datastore, and dedicated worker nodes that share no kernel with any other tenant — aimed at banks, financial institutions, and government.
How does Moncho secure the add-ons it installs?
Add-ons come from a signed catalog — mirrored, scanned, and signed — and are enforced at deploy by an admission policy, so only approved, verified images run in the cluster.
What are Moncho's current limits?
Adding a node joins a pre-staged machine, not elastic autoscale. There is no financially-backed availability SLA until a second-site disaster-recovery drill is published, and a fully in-country substrate for restricted data is planned, not yet deployed.
Adding a node joins a pre-staged machine, not elastic autoscale. There is no financially-backed availability SLA until a second-site disaster-recovery drill is published, and a fully in-country substrate for restricted data is planned, not yet deployed.