Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@agentctl what's the current fleet status?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
K8s-native agent/model workload operator with SAM mesh registration and MCP ops plane.
Replaces process-spawning orchestrators (vllm-orchestrator) with declarative K8s resources. Every operation is exposed as an MCP tool for LangGraph/LangChain workflow composition.
LangGraph agent ──MCP──▶ agentctl ops plane (:8091)
│
┌─────────┼─────────┐
▼ ▼ ▼
workload_* fleet_* sam_*
tools tools tools
│ │ │
▼ ▼ ▼
K8s API K8s API SAM node
(CRDs) (pods) (discovery)Related MCP server: k8s-aiops
pip install agentctl
# or
uv pip install agentctlkubectl apply -f https://github.com/moreWax/agentctl/releases/latest/download/install.yamlThis installs the CRD, the operator, and the MCP ops plane into
agentctl-system. The ops plane is reachable in-cluster at
http://agentctl-mcp.agentctl-system.svc.cluster.local:8091/mcp
(stateless Streamable HTTP).
kubectl apply -f examples/qwen3.8-27b.yaml # Qwen3.8-27B, TP2 on 2 GPUs
kubectl apply -f examples/qwen3.8-27b-fp8.yaml # FP8 variant, single GPU
kubectl apply -f examples/litellm-gateway.yaml # LiteLLM proxy gatewayuv venv && uv pip install -e ".[dev]"
kubectl apply -f crd/agentctl-crd.yaml
agentctl serve # operator (kopf controllers)
agentctl mcp --port 8091 # MCP ops-plane server (stateless Streamable HTTP)kubectl apply -f examples/qwen3.8-27b.yaml # Qwen3.8-27B, TP2 on 2 GPUs
kubectl apply -f examples/qwen3.8-27b-fp8.yaml # FP8 variant, single GPU
kubectl apply -f examples/litellm-gateway.yaml # LiteLLM proxy gatewayThe LiteLLM gateway routes OpenAI-compatible requests to the vLLM backends.
Point your agents at http://litellm-gateway:4000/v1 and use model names like
qwen3.8-27b or qwen3.8-27b-fp8.
The ops plane exposes 3 action-routed tools — a small surface that keeps
agent context cheap and tool selection trivial, while all 20 capabilities
stay available as actions. Everything is stateless, returns structured JSON,
and gates destructive actions behind approved=true.
fleet(action=...) — cluster-wide views (read-only)
Action | Params | Description |
| — | Aggregate: workloads by phase, GPU totals |
| — | Per-node capacity / allocated / free + consumers |
|
| Per-pod: node, boot phase, role, sleeping, GPUs, restarts |
|
| Dry-run placement — call before boot |
|
| Endpoint readiness per service |
|
| SAM mesh discovery |
workload(action=...) — AgentWorkload CRUD
Action | Params | Description |
|
| Workloads with phase, GPUs, pin state |
|
| Full detail incl. per-pod boot phases |
|
| Recent container logs per pod |
|
| 0 parks; refuses pinned scale-down unless forced |
|
| Delete workload + pods |
|
| Create (idempotent) |
|
| Remove |
model(action=...) — model lifecycle & actuation
Action | Params | Description |
|
| Deploy + optional wait → |
|
| Pod phases + vLLM markers ( |
|
| Suspend, free GPUs. FMA: seconds-level ( |
|
| Resume; FMA wake takes seconds |
|
| Force-delete hung pod; Deployment recreates it |
|
| Pin blocks scale-down/sleep unless forced |
Agent boot-wait pattern (LangGraph loop):
model(action="boot", model_id="qwen38-27b", image="vllm/vllm-openai:v0.26.0", gpu_count=2)
while True:
s = model(action="boot_status", model_id="qwen38-27b")
if s["ready"]: break # proceed to inference
if s["failed"]: ... # inspect s["pods"], kill_pod + retry
time.sleep(10) # poll againErrors are structured: {"error": "unknown_action", "valid_actions": [...]},
{"error": "missing_params", "required": [...]}, or {"error": <cause>, "detail": ...}.
uv venv && uv pip install -e ".[dev]"
pytest tests/ -qMIT
This server cannot be deployed
MCP-Native LLM Orchestration Agent
Control plane for autonomous software labor. Agents claim objectives over MCP with audit trail.
MCP Hub: AI service discovery, per-user OAuth, and multi-service workflow orchestration
MCP-first control plane for ProAgentStore agents and private instances.
