Octelium is a free and open source, self-hosted zero trust access platform. It provides identity-based, layer-7 aware access control on a per-request basis, dynamic secretless access, and a unified architecture for both client-based and clientless access, for humans, workloads and AI agents alike.
Humans, workloads and AI agents reach web and internal APIs, SSH and RDP hosts, Kubernetes clusters, databases, MCP servers and LLM providers through one Octelium Cluster. Every request is identified, authorized against the request itself, authenticated to the upstream by the Cluster rather than by the caller, and logged.
Secure humans, workloads and AI agents with one model for identity, authorization, connectivity, credentials and visibility.
One architecture for client-based zero-config VPN-like access over WireGuard and QUIC tunnels and clientless access over browsers for humans and OAuth2 access for workloads.
Read moreIdentity-based, application-layer aware ABAC access control with policy-as-code via CEL and OPA, applied on a per-request basis.
Read moreDynamic secretless access where teams and AI agents access protected infrastructure without ever holding a credential.
Read moreTeams, workloads and AI agents share the same model for identity management, access control, and auditing.
Read moreOpenTelemetry-native, real-time, application-layer-aware visibility and structured logging.
Read moreOctelium is administered like Kubernetes. Define your resources in YAML files, store them in a Git repository.
Read moreEach resource is protected by an identity-aware proxy that provides application-layer-aware access control, secretless access, and visibility. Users reach Services through two equivalent modes.
kind: Service
metadata:
name: grafana
spec:
mode: HTTP
config:
upstream:
url: http://10.0.6.12:3000kind: Service
metadata:
name: grafana
spec:
mode: HTTP
isPublic: true
config:
upstream:
url: http://10.0.6.12:3000kind: Service
metadata:
name: docs
spec:
mode: HTTP
isPublic: true
isAnonymous: true
config:
upstream:
container:
port: 8080
image: ghcr.io/acme/docs:2.1.0Access is authorized on a per-request basis, using CEL or Open Policy Agent, against the identity behind the request and the content of the request itself rather than the network path it arrived on.
kind: Policy
metadata:
name: api-read-only
spec:
rules:
- effect: ALLOW
condition:
all:
of:
- match: ctx.request.http.method == "GET"
- match: ctx.request.http.path == "/v1/stats"
- match: '"analysts" in ctx.user.spec.groups'kind: Policy
metadata:
name: k8s-sre
spec:
rules:
- effect: ALLOW
condition:
all:
of:
- match: ctx.request.kubernetes.verb == "get"
- match: '"sre" in ctx.user.spec.groups'kind: Policy
metadata:
name: sensitive
spec:
rules:
- effect: ALLOW
condition:
all:
of:
- match: ctx.user.spec.type == "HUMAN"
- match: ctx.session.status.type == "CLIENT"
- match: ctx.device.status.osType in ["MAC"]Application-layer credentials are stored in the Cluster as Secrets and injected into the upstream connection once a request is authorized. This removes credential sprawl for teams and, increasingly, for workloads and AI agents.
kind: Service
metadata:
name: pg-prod
spec:
mode: POSTGRES
port: 5432
config:
upstream:
url: postgres://10.0.4.21:5432
postgres:
user: analytics_ro
database: analytics
auth:
password:
fromSecret: pg-password
sslMode: REQUIREkind: Service
metadata:
name: bastion
spec:
mode: SSH
port: 22
config:
upstream:
url: ssh://10.0.2.14
ssh:
user: deploy
auth:
privateKey:
fromSecret: bastion-keykind: Service
metadata:
name: prod-k8s
spec:
mode: KUBERNETES
config:
upstream:
url: https://k8s.internal:6443
kubernetes:
kubeconfig:
fromSecret: kubeconfig-prodHuman and workload Users share the same identity management, authentication, access control and visibility model. For agentic systems this matters more than it once did, because the number of non-human identities holding credentials grows faster than the number of people.
kind: Service
metadata:
name: tools-mcp
spec:
mode: MCP
isPublic: true
config:
upstream:
container:
port: 8080
image: ghcr.io/acme/mcp:1.4.0
replicas: 2
mcp:
endpoint: /mcp
protocol:
requireVersion: truekind: Policy
metadata:
name: mcp-tools
spec:
rules:
- effect: ALLOW
condition:
all:
of:
- match: ctx.request.mcp.method == "tools/call"
- match: ctx.request.mcp.name == "search_docs"
- match: '"agents" in ctx.user.spec.groups'kind: IdentityProvider
metadata:
name: github-actions
spec:
oidcIdentityToken:
issuerURL: https://token.actions.githubusercontent.com
audience: https://example.comOpenTelemetry-ready, application-layer aware, structured auditing and visibility emitted to your OpenTelemetry OTLP receivers, where they can be exported to your log management and SIEM tools.
A Cluster runs on Kubernetes and uses it to scale its own data plane, place Service proxies across Gateways, and run containerized applications that the Cluster itself deploys.
$ octeliumctl apply ./cluster
Secret: pg-password Updated
Secret: bastion-key Updated
IdentityProvider: okta Updated
Policy: analysts Created
Group: analysts Created
Group: sre Updated
Service: pg-prod.db Created
Service: bastion Created
Service: tools-mcp Created
Namespace: db Created
Cluster Core resources successfully applied
7 resources created
3 resources updated
ClusterConfig updatedBoth products are described from the same structured capability dataset used across this site, with the same evidentiary standard applied to each. Emerging and experimental features are marked as partial regardless of vendor.
A curated view of the dataset, including the rows where Octelium is stronger, the rows where it is not, and the rows where the two are equivalent. The complete capability matrix, with search and filters, is one click away.
Architecture
Native FIDO2 / Passkey
Native means implemented by the platform itself, not merely delegated to an external IdP.
Workload Identity
SCIM Provisioning
Automated user/group lifecycle provisioning via SCIM, beyond just login-time IdP federation.
Policy Model
Policy-as-Code
Per-Request Authz
For L7 products, this means each HTTP/gRPC/API/K8s request can be evaluated. For overlays, session/connection policy is not counted as per-request. This modeling choice structurally favors L7 gateways over network overlays.
L7-Aware Policies
Zero Standing Privilege / JIT
Indicates strong architectural support for minimizing standing privilege; not an absolute claim that privileged access cannot exist.
Access Requests / Approvals
First-class request-and-approve / break-glass workflows (reviewers, time-bound grants), as opposed to static policy alone.
Device Posture
SSH
PostgreSQL
Kubernetes
WireGuard Data Plane
Whether the product's own client/data plane is built on WireGuard, versus a proprietary or TLS-based transport.
NAT Traversal
OpenTelemetry-Native
L7-Aware Access Logs
Clientless Browser Access
Workload OAuth2 / Bearer
Private DNS
GitOps / Declarative Config
AI / LLM Gateway
MCP Gateway / MCP Access
Deploy Containers / PaaS
Kubernetes-Native
High Availability
Fully Self-Hosted
Server Code Open Source
No Mandatory Proprietary Cloud
Data Sovereignty
Unified Access Platform
A unified secure access platform with secretless access, CEL/OPA-style ABAC, OpenTelemetry-native real-time visibility, and both client-based and clientless access.
Best for
Tradeoffs
Mesh VPN / Overlay
Best viewed as an identity-aware connectivity mesh rather than an L7 authorization or secretless proxy platform.
Best for
Tradeoffs
The same platform capabilities ship with every deployment, whichever way you use it.
Octelium as a complete ZTNA and BeyondCorp platform for humans, workloads and AI agents on a single architecture.
Read moreZero-config WireGuard tunnels with a single stable route, dual-stack addressing and automatic private DNS.
Read moreIdentity, access control and auditing for AI agents, MCP servers and LLM providers at scale.
Read moreFree and open source. Self-hosted. No vendor lock-in.