Skip to main content
AGENTSCHANNELSTASKSWORKSPACESHERMESOPENCLAWMONITORINGMODELSKUBERNETESVPSCLOUDHYBRID
My Clawster Logo

Own your agents.
Own your stack.

Deploy your first clawster in 5 minutes. No vendor lock-in.
My Clawster is open source.

Star on GitHubJoin Discord
[SELF-HOSTABLE AGENT OPS]

A control room for teams running their own agent clusters.

Self-hostable cluster management for teams running OpenClaw, Hermes, and custom agent frameworks across isolated workspaces.
Deploy OpenClaw, Hermes, and any open-source agent framework — orchestrated together in isolated workspaces. Runs on-premise, cloud, or hybrid. No vendor lock-in. Your data, your rules.

Get your self-hosted AI agent clusters up and running without the ops overhead.
Deploy Your First ClawsterRead the docs
My Clawster Logo
My Clawster cluster overview dashboard
fish
$ clawster status --workspace prodgateway: onlineworkers: 12 isolatedegress: allowlist clean
Latest field note
Clawne Me: Managed AI Agents Built for Humans, Powered by OpenClaw and Hermes

Discover how Clawne Me transforms complex AI agent infrastructure into a simple, human-friendly experience. Learn how managed agents, OpenClaw, Hermes, MCP, and AI workflow automation are redefining personal operational intelligence for non-technical users and enterprises alike.

[WHY MY CLAWSTER]

Bring agent infrastructure in-house without turning ops into a side project.

The best-fit teams want control over isolation, provider choice, and model policy, but they still need a clear story for demos, onboarding, and stakeholder trust. Remove shadow IT barriers by providing teams with a repeatable cluster architecture that supports both single-provider and hybrid deployments, with operational confidence built in.

[OWN THE RUNTIME]

Own the runtime

Keep agent routing, model access, and workspace isolation inside infrastructure your team can inspect and control.

[SHIP ACROSS PROVIDERS]

Ship across providers

Run the same operational model across VPS, Kubernetes, and hybrid footprints without rebuilding your control plane.

[STAY DEMO-READY]

Stay demo-ready

Deploy "Claws" (OpenClaw, or Hermes instances) in a way that supports demos, onboarding, and security reviews without hand-holding.

DEPLOYMENT MODES

Wizard-guided: Single-provider or hybrid.

The onboarding wizard uses architecture previews to explain what changes when you choose one provider path for the whole clawster versus mixing provider types by role.

Single path for every clawHybrid placement by rolePreview-first onboardingBuyer-friendly architecture story
SINGLE PROVIDER · VPS

One VPS fleet for every claw

Fastest path to one repeatable deployment story across the entire clawster.

Diagram of a single-provider VPS clawster with all claw roles running through one VM provider path.

Vultr / General Purpose VPS becomes the one provider path for every claw in the clawster.

Best when you want straightforward host-level control and one VM-style path for every claw.

  • Gateway, worker, and support claws all follow the same VM-style path.
  • Useful for teams that want straightforward host-level control.
  • Examples: Contabo, Hetzner, Vultr, or similar VPS offerings.
HYBRID

Mix provider types by claw role

Best when demos, workers, gateways, and managed integrations should not share one provider constraint.

Diagram of a hybrid clawster with different claw roles distributed across VPS, Kubernetes, and cloud provider paths.

Start with one primary provider binding, then place specific claw roles where they fit best, such as VPS at the edge, Kubernetes for workers, and cloud services for managed integrations.

Hybrid is the right fit when one provider path is not ideal for every claw role or lifecycle boundary.

  • Keep a shared clawster identity while splitting roles across provider types.
  • Useful for patterns like VPS gateways plus Kubernetes workers.
  • Best when one provider path is not ideal for every role.
DEPLOY AGENTS•CONNECT CHANNELS•ROUTE TASKS•ISOLATE WORKSPACES•MANAGE CREDENTIALS•MONITOR COSTS•SCHEDULE CRON•CONFIGURE MODELS•DEPLOY AGENTS•CONNECT CHANNELS•ROUTE TASKS•ISOLATE WORKSPACES•MANAGE CREDENTIALS•MONITOR COSTS•SCHEDULE CRON•CONFIGURE MODELS•
WHATSAPP•TELEGRAM•DISCORD•SLACK•SIGNAL•iMESSAGE•MATRIX•TEAMS•WEBCHAT•IRC•NOSTR•LINE•FEISHU•WHATSAPP•TELEGRAM•DISCORD•SLACK•SIGNAL•iMESSAGE•MATRIX•TEAMS•WEBCHAT•IRC•NOSTR•LINE•FEISHU•
CLAUDE•GPT-4o•GEMINI•DEEPSEEK•LLAMA 4•MINIMAX•OLLAMA•OPENROUTER•CUSTOM ENDPOINTS•CLAUDE•GPT-4o•GEMINI•DEEPSEEK•LLAMA 4•MINIMAX•OLLAMA•OPENROUTER•CUSTOM ENDPOINTS•
SELF-IMPROVING•EPISODIC MEMORY•PERSISTENT SESSIONS•HEARTBEAT SCHEDULING•CONTAINER ISOLATION•AUDIT TRAILS•GIT VERSIONING•COST DASHBOARDS•SELF-IMPROVING•EPISODIC MEMORY•PERSISTENT SESSIONS•HEARTBEAT SCHEDULING•CONTAINER ISOLATION•AUDIT TRAILS•GIT VERSIONING•COST DASHBOARDS•
[OPERATING MODEL]

Repeatable cluster architecture

A reusable cluster layer designed to support agent frameworks, policy boundaries, and provider mobility, without surrendering ownership of the runtime that supports provider flexibility and clear isolation.

Workspace isolation

Separate customer, project, or environment contexts without giving up cluster-level visibility.

Model routing

Blend Ollama, hosted APIs, MCPs, and BYOK policies with one operational story instead of fragmented scripts.

Provider mobility

Start a VPS today, move into Kubernetes tomorrow, and keep the workflows recognizable for your team.

[CLUSTER SIGNAL]

What DevSecOps need to see fast

  • Clear isolation story for enterprise needs.
  • Provider flexibility without forcing teams into bespoke tooling.
  • Operational confidence that the clawster's security posture and runtime health are visible and verifiable.
[LATEST]

Operational notes from the cluster edge

Architecture decisions, provider notes, deployment patterns, and product thinking for teams that want agent infrastructure without vendor handcuffs.

Clawne MeOpenClawHermesManaged AgentsAI Assistants
EssayApr 25, 20261 min

posts

Field notes, launch updates, and deployment writeups for My Clawster.

[UPDATES]

Stay close to the cluster

Get release notes, architecture guides, and field-tested patterns for self-hosted agent infrastructure.

[COMMUNITY]

Built by builders who run their own agents

Open source, MIT licensed — your agents, your data, your rules.

0K+
OpenClaw Stars
0K+
Hermes Stars
0+
Channels
0+
Models

247 clusters deployed and counting