diff --git a/docs/.vitepress/theme/components/Home.vue b/docs/.vitepress/theme/components/Home.vue index a3603a77..1101bd54 100644 --- a/docs/.vitepress/theme/components/Home.vue +++ b/docs/.vitepress/theme/components/Home.vue @@ -57,9 +57,9 @@ onUnmounted(() => {
-

The Enterprise Model Registry
for Secure AI

+

The Enterprise Registry for
Models, Agents, and MCP Servers

Secure. Portable. No Lock‑In.

-

The only model registry that works anywhere containers run — from public cloud to the most locked‑down air‑gapped environments.

+

The only AI registry that packages and governs models, agent skills, and MCP servers as OCI artifacts — from public cloud to the most locked‑down air-gapped environments.

@@ -139,8 +139,8 @@ onUnmounted(() => {

Why Kitops exists

-

Enterprise AI teams face three problems with traditional model registries: security gaps, vendor lock‑in, and deployment friction.
KitOps solves them all.

-

Unlike proprietary registries that trap you in closed ecosystems, KitOps stores models, datasets, code, and configs as OCI artifacts in your container registries — Docker Hub, ECR, GCR, Harbor, Artifactory, and beyond.

+

Enterprise AI is no longer just models. Teams are deploying agents, MCP servers, skills, and policies — all of which need the same versioning, security, and governance as container images. KitOps packages them all as standard OCI artifacts in the registries you already run.

+

From proprietary repos that trap your AI artifacts in closed ecosystems to fragile ad‑hoc approaches that bypass your security toolchain — KitOps brings models, agent skills, MCP servers, datasets, and configuration under one governed workflow.

With 100,000+ downloads and 18+ months in production, KitOps is trusted by organizations that can’t compromise on security, compliance, or uptime.

@@ -169,34 +169,18 @@ onUnmounted(() => {
-
Security That’s Built In — Not Bolted On
-
    -
  • Immutable by design — once published, artifacts can’t be tampered with
    • -
  • Cryptographically signed — every model, dataset, and config is verifiable
    • -
  • Provenance tracking — full supply chain auditability out of the box
    • -
  • Defense in depth — layered SHA digest verification for all artifacts
    • -
+
One Package for Your Entire AI Stack
+

Bundle models, datasets, code, agent skills, MCP servers, guardrail configs, and policies into a single versioned OCI artifact. No more scattered repos, zip files, or git submodules for your AI components.

-
Fits Into Your Toolchain — No Re‑Platforming
-
    -
  • OCI‑native — runs on any compliant container registry
    • -
  • No proprietary formats — open standards mean zero lock‑in
    • -
  • CI/CD‑ready — GitHub Actions, GitLab CI, Jenkins, Azure DevOps, and more
    • -
  • Deployment‑friendly — works with Kubernetes, Docker, and existing pipelines
    • -
+
Supply Chain Security, Built In
+

Sign with Cosign. Scan before promotion. Generate SBOMs. Every component is SHA‑256 hashed and tamper‑evident — the same security posture you apply to container images, now applied to your agents and MCP servers.

-
Enterprise‑Grade from Day One
-
    -
  • Air‑gapped deployments — fully supported
    • -
  • Multi‑cloud portability — same artifact runs anywhere
    • -
  • Deduplicated storage — cuts registry costs by up to 80%
    • -
  • Automatic containerization — for any model
    • -
  • RBAC integration — aligns with your security policies
    • -
+
Your Registry. Your Rules.
+

Store ModelKits in Docker Hub, ECR, GCR, Harbor, Artifactory, or Jozu Hub. No proprietary registry required. Your existing access controls, scanning, and compliance workflows just work.

@@ -303,29 +287,21 @@ onUnmounted(() => {
-
Momentum Focused
-
    -
  • Trusted in production
    • -
  • Relied on by Enterprise teams
    • -
  • Engineered for adoption
    • -
+
Immutable Packages
+

Every model, agent skill, and MCP server is a versioned, immutable OCI artifact. What you deployed is exactly what was tested.

-
Trust & Security Focused
-
    -
  • Built to meet proven standards
    • -
  • Designed for air-gapped environments
    • -
  • Security by default
    • -
+
Content-Addressable Storage
+

SHA-256 digests for every component. Tamper detection works the same way it does for your container images — because it is the same infrastructure.

-
Integration & Workflow Focused
+
Designed for Air Gap
    +
  • Designed for air-gapped environments
    • +
  • Security by default
  • Native to your stack
    • -
  • Works at enterprise scale
    • -
  • Moves at CI/CD speed
@@ -341,42 +317,42 @@ onUnmounted(() => {
- Air-gapped deployments + Cryptographic Security
- Fully supported + Sign and verify models, agent skills, and MCP servers with the same Cosign workflow used for container images.
- Multi-cloud portability + Full Version Packaging
- Same artifact runs everywhere + Package your complete AI stack — models, skills, MCP servers, policies — as a single immutable OCI artifact with selective pull for any component.
- Deduplicated storage + Designed for Air Gap
- Cuts registry costs by up to 80% + Fully supported
@@ -385,21 +361,21 @@ onUnmounted(() => {
- Automatic containerization + Automated Compliance
- For any model + Generate AI Bill of Materials covering models, agent skills, and MCP tool dependencies. Map directly to EU AI Act and NIST AI RMF requirements.
- RBAC integration + Artifact Tracking
@@ -437,7 +413,8 @@ onUnmounted(() => {
-

Package Your First Model

+

Package Your First Agent

+

Bundle a model, MCP server config, and agent skills into a signed ModelKit in under 5 minutes.

kit pack . -t myregistry.com/my-model:latest
@@ -548,6 +525,24 @@ onUnmounted(() => {

Enterprise support for ModelKits and the Kit CLI is available from Jozu.

+ + + + +

Yes. KitOps packages any AI/ML component — models, datasets, code, configuration, MCP server definitions, agent skills, guardrail configs, and policies — as a single versioned OCI artifact. MCP servers and agent skills get the same cryptographic signing, scanning, and governance as your models.

+ + + + + +

OCI artifacts give you immutable versioning, cryptographic signatures, supply chain scanning, and content-addressable storage — the same infrastructure you already use for container images. Git repos work for source code, but production AI deployments need the tamper-evidence, provenance, and policy gating that OCI provides.

+ + + + + +

KitOps is the reference implementation of ModelPack, the CNCF specification for vendor-neutral AI/ML packaging. KitOps also supports its own ModelKit format with broader artifact coverage including MCP servers, agent skills, and policies.

+