Add unikernel runtime: run services as Nanos VMs under QEMU/KVM

Implements the hypervisor design's Phase 1: a second runtime.Runtime backend (QEMU) that runs each service component as a Nanos unikernel VM instead of a podman container, selected per-component via a new runtime = "unikernel" service-def field. - internal/runtime/qemu.go: QEMURuntime. Pull extracts the ELF from the OCI image; Run does `ops build` + boots qemu-system-x86_64 with KVM, user-mode net port-forwards, QMP control socket and serial console log; Stop/Remove/Inspect/List/Logs map onto VM lifecycle + state dir. - proto/registry/servicedef: add runtime, memory_mb, vcpus fields (registry migration 5). - agent: holds both runtimes; runtimeFor() selects per component; listAllContainers() merges containers + VMs so drift/status see both. Unikernel runtime auto-enables on nodes with /dev/kvm + ops. Validated end-to-end on straylight: a test service deploys via `mcp deploy --direct`, boots as a Nanos unikernel, serves HTTP through the agent port-forward, and reports running via `mcp status`/`mcp logs`. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-11 00:54:49 -07:00
parent 3b08caaa0a
commit d56f224359
30 changed files with 949 additions and 152 deletions
--- a/internal/agent/adopt.go
+++ b/internal/agent/adopt.go
@@ -19,7 +19,7 @@ func (a *Agent) AdoptContainers(ctx context.Context, req *mcpv1.AdoptContainersR
 		return nil, fmt.Errorf("service name is required")
 	}

-	containers, err := a.Runtime.List(ctx)
+	containers, err := a.listAllContainers(ctx)
 	if err != nil {
 		return nil, fmt.Errorf("list containers: %w", err)
 	}
@@ -46,7 +46,7 @@ func (a *Agent) AdoptContainers(ctx context.Context, req *mcpv1.AdoptContainersR
 	}

 	// Ensure the service exists once, before adopting any containers.
-	if err := registry.CreateService(a.DB, service, true); err != nil {
+	if err := registry.CreateService(a.DB, service, true, ""); err != nil {
 		if _, getErr := registry.GetService(a.DB, service); getErr != nil {
 			return nil, fmt.Errorf("create service %q: %w", service, err)
 		}
--- a/internal/agent/agent.go
+++ b/internal/agent/agent.go
@@ -9,6 +9,7 @@ import (
 	"net"
 	"os"
 	"os/signal"
+	"path/filepath"
 	"syscall"

 	mcpv1 "git.wntrmute.dev/mc/mcp/gen/mcp/v1"
@@ -28,7 +29,8 @@ type Agent struct {

 	Config    *config.AgentConfig
 	DB        *sql.DB
-	Runtime   runtime.Runtime
+	Runtime   runtime.Runtime // container runtime (podman)
+	Unikernel runtime.Runtime // unikernel runtime (qemu/nanos); nil if unavailable
 	Monitor   *monitor.Monitor
 	Logger    *slog.Logger
 	PortAlloc *PortAllocator
@@ -53,6 +55,18 @@ func Run(cfg *config.AgentConfig, version string) error {

 	rt := &runtime.Podman{}

+	// The unikernel runtime is enabled only on nodes with KVM. Services with
+	// runtime = "unikernel" are placed by the master on KVM-capable nodes.
+	var uk runtime.Runtime
+	if unikernelSupported() {
+		uk = &runtime.QEMU{
+			ImageDir: filepath.Join(homeDir(cfg), "images"),
+			StateDir: filepath.Join(homeDir(cfg), "vm"),
+			HomeDir:  homeDir(cfg),
+		}
+		logger.Info("unikernel runtime enabled (KVM detected)")
+	}
+
 	mon := monitor.New(db, rt, cfg.Monitor, cfg.Agent.NodeName, logger)

 	proxy, err := NewProxyRouter(cfg.MCProxy.Socket, cfg.MCProxy.CertDir, logger)
@@ -74,6 +88,7 @@ func Run(cfg *config.AgentConfig, version string) error {
 		Config:    cfg,
 		DB:        db,
 		Runtime:   rt,
+		Unikernel: uk,
 		Monitor:   mon,
 		Logger:    logger,
 		PortAlloc: NewPortAllocator(),
--- a/internal/agent/boot.go
+++ b/internal/agent/boot.go
@@ -120,7 +120,7 @@ func (a *Agent) checkServiceHealth(ctx context.Context, serviceName, method stri
 // or from the running container.
 func (a *Agent) findServicePort(serviceName string) (int, error) {
 	// Check the running containers for a mapped port.
-	containers, err := a.Runtime.List(context.Background())
+	containers, err := a.listAllContainers(context.Background())
 	if err != nil {
 		return 0, fmt.Errorf("list containers: %w", err)
 	}
--- a/internal/agent/deploy.go
+++ b/internal/agent/deploy.go
@@ -22,7 +22,7 @@ func (a *Agent) Deploy(ctx context.Context, req *mcpv1.DeployRequest) (*mcpv1.De
 	serviceName := spec.GetName()
 	a.Logger.Info("deploying", "service", serviceName)

-	if err := ensureService(a.DB, serviceName, spec.GetActive()); err != nil {
+	if err := ensureService(a.DB, serviceName, spec.GetActive(), spec.GetComment()); err != nil {
 		return nil, fmt.Errorf("deploy: ensure service %q: %w", serviceName, err)
 	}

@@ -94,8 +94,20 @@ func (a *Agent) deployComponent(ctx context.Context, serviceName string, cs *mcp
 		Volumes:      cs.GetVolumes(),
 		Cmd:          cs.GetCmd(),
 		Routes:       regRoutes,
+		Runtime:      cs.GetRuntime(),
+		MemoryMB:     int(cs.GetMemoryMb()),
+		VCPUs:        int(cs.GetVcpus()),
 	}

+	// Select the runtime backend (container vs unikernel) for this component.
+	if cs.GetRuntime() == "unikernel" && a.Unikernel == nil {
+		return &mcpv1.ComponentResult{
+			Name:  compName,
+			Error: "service requests unikernel runtime but this node has no KVM/ops support",
+		}
+	}
+	rt := a.runtimeFor(cs.GetRuntime())
+
 	if err := ensureComponent(a.DB, regComp); err != nil {
 		return &mcpv1.ComponentResult{
 			Name:  compName,
@@ -103,27 +115,29 @@ func (a *Agent) deployComponent(ctx context.Context, serviceName string, cs *mcp
 		}
 	}

-	if err := a.Runtime.Pull(ctx, cs.GetImage()); err != nil {
+	if err := rt.Pull(ctx, cs.GetImage()); err != nil {
 		return &mcpv1.ComponentResult{
 			Name:  compName,
 			Error: fmt.Sprintf("pull image: %v", err),
 		}
 	}

-	_ = a.Runtime.Stop(ctx, containerName)   // may not exist yet
-	_ = a.Runtime.Remove(ctx, containerName) // may not exist yet
+	_ = rt.Stop(ctx, containerName)   // may not exist yet
+	_ = rt.Remove(ctx, containerName) // may not exist yet

 	// Build the container spec. If the component has routes, use route-based
 	// port allocation and env injection. Otherwise, fall back to legacy ports.
 	runSpec := runtime.ContainerSpec{
-		Name:    containerName,
-		Image:   cs.GetImage(),
-		Network: cs.GetNetwork(),
-		User:    cs.GetUser(),
-		Restart: cs.GetRestart(),
-		Volumes: cs.GetVolumes(),
-		Cmd:     cs.GetCmd(),
-		Env:     cs.GetEnv(),
+		Name:     containerName,
+		Image:    cs.GetImage(),
+		Network:  cs.GetNetwork(),
+		User:     cs.GetUser(),
+		Restart:  cs.GetRestart(),
+		Volumes:  cs.GetVolumes(),
+		Cmd:      cs.GetCmd(),
+		Env:      cs.GetEnv(),
+		MemoryMB: int(cs.GetMemoryMb()),
+		VCPUs:    int(cs.GetVcpus()),
 	}

 	if len(regRoutes) > 0 && a.PortAlloc != nil {
@@ -142,7 +156,7 @@ func (a *Agent) deployComponent(ctx context.Context, serviceName string, cs *mcp
 		runSpec.Ports = cs.GetPorts()
 	}

-	if err := a.Runtime.Run(ctx, runSpec); err != nil {
+	if err := rt.Run(ctx, runSpec); err != nil {
 		_ = registry.UpdateComponentState(a.DB, serviceName, compName, "", "removed")
 		return &mcpv1.ComponentResult{
 			Name:  compName,
@@ -219,16 +233,16 @@ func (a *Agent) allocateRoutePorts(service, component string, routes []registry.
 }

 // ensureService creates the service if it does not exist, or updates its
-// active flag if it does.
-func ensureService(db *sql.DB, name string, active bool) error {
+// active flag and comment if it does.
+func ensureService(db *sql.DB, name string, active bool, comment string) error {
 	_, err := registry.GetService(db, name)
 	if errors.Is(err, sql.ErrNoRows) {
-		return registry.CreateService(db, name, active)
+		return registry.CreateService(db, name, active, comment)
 	}
 	if err != nil {
 		return err
 	}
-	return registry.UpdateServiceActive(db, name, active)
+	return registry.UpdateServiceActive(db, name, active, comment)
 }

 // hasL7Routes reports whether any route uses L7 (TLS-terminating) mode.
--- a/internal/agent/deploy_test.go
+++ b/internal/agent/deploy_test.go
@@ -34,7 +34,7 @@ func testAgent(t *testing.T) *Agent {
 // allocateRoutePorts can store host ports for it.
 func seedComponent(t *testing.T, db *sql.DB, service, component string, routes []registry.Route) {
 	t.Helper()
-	if err := registry.CreateService(db, service, true); err != nil {
+	if err := registry.CreateService(db, service, true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(db, &registry.Component{
--- a/internal/agent/edge_rpc.go
+++ b/internal/agent/edge_rpc.go
@@ -12,8 +12,8 @@ import (
 	"google.golang.org/grpc/codes"
 	"google.golang.org/grpc/status"

-	mcpv1 "git.wntrmute.dev/mc/mcp/gen/mcp/v1"
 	mcproxy "git.wntrmute.dev/mc/mc-proxy/client/mcproxy"
+	mcpv1 "git.wntrmute.dev/mc/mcp/gen/mcp/v1"
 	"git.wntrmute.dev/mc/mcp/internal/registry"
 )

@@ -182,7 +182,7 @@ func (a *Agent) HealthCheck(_ context.Context, _ *mcpv1.HealthCheckRequest) (*mc

 	// Count running containers if the runtime is available.
 	if a.Runtime != nil {
-		if list, err := a.Runtime.List(context.Background()); err == nil {
+		if list, err := a.listAllContainers(context.Background()); err == nil {
 			containers = int32(len(list)) //nolint:gosec // container count is small
 		} else {
 			st = "degraded"
--- a/internal/agent/heartbeat.go
+++ b/internal/agent/heartbeat.go
@@ -21,8 +21,8 @@ import (
 // MasterConfig holds the optional master connection settings for the agent.
 // When configured, the agent self-registers and sends periodic heartbeats.
 type MasterConfig struct {
-	Address   string `toml:"address"`   // master gRPC address
-	CACert    string `toml:"ca_cert"`   // CA cert to verify master's TLS
+	Address   string `toml:"address"`    // master gRPC address
+	CACert    string `toml:"ca_cert"`    // CA cert to verify master's TLS
 	TokenPath string `toml:"token_path"` // MCIAS service token for auth
 }

@@ -38,12 +38,20 @@ type HeartbeatClient struct {
 	interval time.Duration
 	stop     chan struct{}
 	wg       sync.WaitGroup
-	logger   interface{ Info(string, ...any); Warn(string, ...any); Error(string, ...any) }
+	logger   interface {
+		Info(string, ...any)
+		Warn(string, ...any)
+		Error(string, ...any)
+	}
 }

 // NewHeartbeatClient creates a client that registers with the master and
 // sends periodic heartbeats. Returns nil if master address is not configured.
-func NewHeartbeatClient(cfg config.AgentConfig, logger interface{ Info(string, ...any); Warn(string, ...any); Error(string, ...any) }) (*HeartbeatClient, error) {
+func NewHeartbeatClient(cfg config.AgentConfig, logger interface {
+	Info(string, ...any)
+	Warn(string, ...any)
+	Error(string, ...any)
+}) (*HeartbeatClient, error) {
 	// Config takes precedence, env vars as fallback.
 	masterAddr := cfg.Master.Address
 	if masterAddr == "" {
--- a/internal/agent/lifecycle.go
+++ b/internal/agent/lifecycle.go
@@ -51,7 +51,7 @@ func (a *Agent) StopService(ctx context.Context, req *mcpv1.StopServiceRequest)
 			}
 		}

-		if err := a.Runtime.Stop(ctx, containerName); err != nil {
+		if err := a.runtimeFor(c.Runtime).Stop(ctx, containerName); err != nil {
 			a.Logger.Info("stop container (ignored)", "container", containerName, "error", err)
 		}

@@ -132,12 +132,14 @@ func startComponent(ctx context.Context, a *Agent, service string, c *registry.C
 	containerName := ContainerNameFor(service, c.Name)
 	r := &mcpv1.ComponentResult{Name: c.Name, Success: true}

+	rt := a.runtimeFor(c.Runtime)
+
 	// Remove any pre-existing container; ignore errors for non-existent ones.
-	_ = a.Runtime.Stop(ctx, containerName)
-	_ = a.Runtime.Remove(ctx, containerName)
+	_ = rt.Stop(ctx, containerName)
+	_ = rt.Remove(ctx, containerName)

 	spec := componentToSpec(service, c)
-	if err := a.Runtime.Run(ctx, spec); err != nil {
+	if err := rt.Run(ctx, spec); err != nil {
 		r.Success = false
 		r.Error = fmt.Sprintf("run container: %v", err)
 		return r
@@ -156,11 +158,12 @@ func restartComponent(ctx context.Context, a *Agent, service string, c *registry
 	containerName := ContainerNameFor(service, c.Name)
 	r := &mcpv1.ComponentResult{Name: c.Name, Success: true}

-	_ = a.Runtime.Stop(ctx, containerName)
-	_ = a.Runtime.Remove(ctx, containerName)
+	rt := a.runtimeFor(c.Runtime)
+	_ = rt.Stop(ctx, containerName)
+	_ = rt.Remove(ctx, containerName)

 	spec := componentToSpec(service, c)
-	if err := a.Runtime.Run(ctx, spec); err != nil {
+	if err := rt.Run(ctx, spec); err != nil {
 		r.Success = false
 		r.Error = fmt.Sprintf("run container: %v", err)
 		_ = registry.UpdateComponentState(a.DB, service, c.Name, "", "stopped")
@@ -177,14 +180,16 @@ func restartComponent(ctx context.Context, a *Agent, service string, c *registry
 // componentToSpec builds a runtime.ContainerSpec from a registry Component.
 func componentToSpec(service string, c *registry.Component) runtime.ContainerSpec {
 	return runtime.ContainerSpec{
-		Name:    ContainerNameFor(service, c.Name),
-		Image:   c.Image,
-		Network: c.Network,
-		User:    c.UserSpec,
-		Restart: c.Restart,
-		Ports:   c.Ports,
-		Volumes: c.Volumes,
-		Cmd:     c.Cmd,
+		Name:     ContainerNameFor(service, c.Name),
+		Image:    c.Image,
+		Network:  c.Network,
+		User:     c.UserSpec,
+		Restart:  c.Restart,
+		Ports:    c.Ports,
+		Volumes:  c.Volumes,
+		Cmd:      c.Cmd,
+		MemoryMB: c.MemoryMB,
+		VCPUs:    c.VCPUs,
 	}
 }

--- a/internal/agent/logs.go
+++ b/internal/agent/logs.go
@@ -2,15 +2,22 @@ package agent

 import (
 	"bufio"
+	"context"
 	"io"
+	"os/exec"

 	mcpv1 "git.wntrmute.dev/mc/mcp/gen/mcp/v1"
 	"git.wntrmute.dev/mc/mcp/internal/registry"
-	"git.wntrmute.dev/mc/mcp/internal/runtime"
 	"google.golang.org/grpc/codes"
 	"google.golang.org/grpc/status"
 )

+// logStreamer is implemented by both the Podman (container logs) and QEMU
+// (serial console) runtimes.
+type logStreamer interface {
+	Logs(ctx context.Context, name string, tail int, follow, timestamps bool, since string) *exec.Cmd
+}
+
 // Logs streams container logs for a service component.
 func (a *Agent) Logs(req *mcpv1.LogsRequest, stream mcpv1.McpAgentService_LogsServer) error {
 	if req.GetService() == "" {
@@ -32,14 +39,20 @@ func (a *Agent) Logs(req *mcpv1.LogsRequest, stream mcpv1.McpAgentService_LogsSe

 	containerName := ContainerNameFor(req.GetService(), component)

-	podman, ok := a.Runtime.(*runtime.Podman)
+	// Select the runtime for this component (container vs unikernel) and
+	// stream its logs (podman logs / journald, or the VM serial console).
+	var compRuntime string
+	if c, err := registry.GetComponent(a.DB, req.GetService(), component); err == nil {
+		compRuntime = c.Runtime
+	}
+	ls, ok := a.runtimeFor(compRuntime).(logStreamer)
 	if !ok {
-		return status.Error(codes.Internal, "logs requires podman runtime")
+		return status.Error(codes.Internal, "selected runtime does not support log streaming")
 	}

-	cmd := podman.Logs(stream.Context(), containerName, int(req.GetTail()), req.GetFollow(), req.GetTimestamps(), req.GetSince())
+	cmd := ls.Logs(stream.Context(), containerName, int(req.GetTail()), req.GetFollow(), req.GetTimestamps(), req.GetSince())

-	a.Logger.Info("running podman logs", "container", containerName, "args", cmd.Args)
+	a.Logger.Info("streaming logs", "container", containerName, "runtime", compRuntime, "args", cmd.Args)

 	// Podman writes container stdout to its stdout and container stderr
 	// to its stderr. Merge both into a single pipe.
@@ -48,7 +61,7 @@ func (a *Agent) Logs(req *mcpv1.LogsRequest, stream mcpv1.McpAgentService_LogsSe
 	cmd.Stderr = pw

 	if err := cmd.Start(); err != nil {
-		pw.Close()
+		_ = pw.Close()
 		return status.Errorf(codes.Internal, "start podman logs: %v", err)
 	}

@@ -58,7 +71,7 @@ func (a *Agent) Logs(req *mcpv1.LogsRequest, stream mcpv1.McpAgentService_LogsSe
 		if err != nil {
 			a.Logger.Warn("podman logs exited", "container", containerName, "error", err)
 		}
-		pw.Close()
+		_ = pw.Close()
 	}()

 	scanner := bufio.NewScanner(pr)
--- a/internal/agent/purge_test.go
+++ b/internal/agent/purge_test.go
@@ -14,7 +14,7 @@ func TestPurgeComponentRemoved(t *testing.T) {
 	ctx := context.Background()

 	// Set up a service with a stale component.
-	if err := registry.CreateService(a.DB, "mcns", true); err != nil {
+	if err := registry.CreateService(a.DB, "mcns", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -69,7 +69,7 @@ func TestPurgeRefusesRunning(t *testing.T) {
 	a := newTestAgent(t, rt)
 	ctx := context.Background()

-	if err := registry.CreateService(a.DB, "mcr", true); err != nil {
+	if err := registry.CreateService(a.DB, "mcr", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -109,7 +109,7 @@ func TestPurgeRefusesStopped(t *testing.T) {
 	a := newTestAgent(t, rt)
 	ctx := context.Background()

-	if err := registry.CreateService(a.DB, "mcr", true); err != nil {
+	if err := registry.CreateService(a.DB, "mcr", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -140,7 +140,7 @@ func TestPurgeSkipsDefinedComponent(t *testing.T) {
 	a := newTestAgent(t, rt)
 	ctx := context.Background()

-	if err := registry.CreateService(a.DB, "mcns", true); err != nil {
+	if err := registry.CreateService(a.DB, "mcns", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -176,7 +176,7 @@ func TestPurgeDryRun(t *testing.T) {
 	a := newTestAgent(t, rt)
 	ctx := context.Background()

-	if err := registry.CreateService(a.DB, "mcns", true); err != nil {
+	if err := registry.CreateService(a.DB, "mcns", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -217,7 +217,7 @@ func TestPurgeServiceFilter(t *testing.T) {
 	ctx := context.Background()

 	// Create two services.
-	if err := registry.CreateService(a.DB, "mcns", true); err != nil {
+	if err := registry.CreateService(a.DB, "mcns", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -226,7 +226,7 @@ func TestPurgeServiceFilter(t *testing.T) {
 	}); err != nil {
 		t.Fatalf("create component: %v", err)
 	}
-	if err := registry.CreateService(a.DB, "mcr", true); err != nil {
+	if err := registry.CreateService(a.DB, "mcr", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -263,7 +263,7 @@ func TestPurgeServiceDeletedWhenEmpty(t *testing.T) {
 	a := newTestAgent(t, rt)
 	ctx := context.Background()

-	if err := registry.CreateService(a.DB, "mcns", true); err != nil {
+	if err := registry.CreateService(a.DB, "mcns", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -306,7 +306,7 @@ func TestPurgeServiceKeptWhenComponentsRemain(t *testing.T) {
 	a := newTestAgent(t, rt)
 	ctx := context.Background()

-	if err := registry.CreateService(a.DB, "mcns", true); err != nil {
+	if err := registry.CreateService(a.DB, "mcns", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	// Stale component (will be purged).
@@ -359,7 +359,7 @@ func TestPurgeExitedState(t *testing.T) {
 	a := newTestAgent(t, rt)
 	ctx := context.Background()

-	if err := registry.CreateService(a.DB, "test", true); err != nil {
+	if err := registry.CreateService(a.DB, "test", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -384,7 +384,7 @@ func TestPurgeUnknownState(t *testing.T) {
 	a := newTestAgent(t, rt)
 	ctx := context.Background()

-	if err := registry.CreateService(a.DB, "test", true); err != nil {
+	if err := registry.CreateService(a.DB, "test", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
--- a/internal/agent/recover.go
+++ b/internal/agent/recover.go
@@ -22,7 +22,7 @@ func (a *Agent) Recover(ctx context.Context) error {
 	}

 	// Get the list of currently running containers from podman.
-	running, err := a.Runtime.List(ctx)
+	running, err := a.listAllContainers(ctx)
 	if err != nil {
 		a.Logger.Warn("cannot list containers, assuming none running", "err", err)
 		running = nil
@@ -67,18 +67,22 @@ func (a *Agent) Recover(ctx context.Context) error {
 				"image", comp.Image,
 			)

+			rt := a.runtimeFor(comp.Runtime)
+
 			// Remove any stale container with the same name.
-			_ = a.Runtime.Remove(ctx, containerName)
+			_ = rt.Remove(ctx, containerName)

 			// Build the container spec from the registry.
 			spec := runtime.ContainerSpec{
-				Name:    containerName,
-				Image:   comp.Image,
-				Network: comp.Network,
-				User:    comp.UserSpec,
-				Restart: comp.Restart,
-				Volumes: comp.Volumes,
-				Cmd:     comp.Cmd,
+				Name:     containerName,
+				Image:    comp.Image,
+				Network:  comp.Network,
+				User:     comp.UserSpec,
+				Restart:  comp.Restart,
+				Volumes:  comp.Volumes,
+				Cmd:      comp.Cmd,
+				MemoryMB: comp.MemoryMB,
+				VCPUs:    comp.VCPUs,
 			}

 			// Allocate ports from routes if the component has routes.
@@ -95,7 +99,7 @@ func (a *Agent) Recover(ctx context.Context) error {
 				spec.Ports = comp.Ports
 			}

-			if err := a.Runtime.Run(ctx, spec); err != nil {
+			if err := rt.Run(ctx, spec); err != nil {
 				a.Logger.Error("recover container failed",
 					"container", containerName,
 					"err", err,
--- a/internal/agent/runtime.go
+++ b/internal/agent/runtime.go
@@ -0,0 +1,65 @@
+package agent
+
+import (
+	"context"
+	"os"
+	"os/exec"
+	"path/filepath"
+
+	"git.wntrmute.dev/mc/mcp/internal/config"
+	"git.wntrmute.dev/mc/mcp/internal/runtime"
+)
+
+// unikernelSupported reports whether this node can run Nanos unikernels:
+// it needs KVM (/dev/kvm) and the `ops` toolchain on PATH.
+func unikernelSupported() bool {
+	if _, err := os.Stat("/dev/kvm"); err != nil {
+		return false
+	}
+	if _, err := exec.LookPath("ops"); err != nil {
+		return false
+	}
+	return true
+}
+
+// homeDir returns the agent's working directory (where images/ and vm/ live),
+// derived from the registry database path (e.g. /srv/mcp/mcp.db -> /srv/mcp).
+func homeDir(cfg *config.AgentConfig) string {
+	if cfg != nil && cfg.Database.Path != "" {
+		return filepath.Dir(cfg.Database.Path)
+	}
+	if h := os.Getenv("HOME"); h != "" {
+		return h
+	}
+	return "/srv/mcp"
+}
+
+// runtimeFor selects the runtime backend for a component's declared runtime.
+// Unknown or empty runtimes fall back to the container runtime. If a service
+// requests "unikernel" but this node lacks the unikernel runtime, it falls
+// back to the container runtime (the master should not place it here).
+func (a *Agent) runtimeFor(rt string) runtime.Runtime {
+	if rt == "unikernel" && a.Unikernel != nil {
+		return a.Unikernel
+	}
+	return a.Runtime
+}
+
+// listAllContainers returns the observed state across every configured
+// runtime (containers + unikernel VMs) so reconciliation, status, and drift
+// detection see the whole picture.
+func (a *Agent) listAllContainers(ctx context.Context) ([]runtime.ContainerInfo, error) {
+	infos, err := a.Runtime.List(ctx)
+	if err != nil {
+		return nil, err
+	}
+	if a.Unikernel != nil {
+		vms, vmErr := a.Unikernel.List(ctx)
+		if vmErr == nil {
+			infos = append(infos, vms...)
+		} else if a.Logger != nil {
+			a.Logger.Warn("list unikernel VMs failed", "err", vmErr)
+		}
+	}
+	return infos, nil
+}
--- a/internal/agent/status.go
+++ b/internal/agent/status.go
@@ -15,8 +15,9 @@ import (
 // ServiceInfo message.
 func buildServiceInfo(svc registry.Service, components []registry.Component) *mcpv1.ServiceInfo {
 	info := &mcpv1.ServiceInfo{
-		Name:   svc.Name,
-		Active: svc.Active,
+		Name:    svc.Name,
+		Active:  svc.Active,
+		Comment: svc.Comment,
 	}
 	for _, c := range components {
 		info.Components = append(info.Components, &mcpv1.ComponentInfo{
@@ -56,7 +57,7 @@ func (a *Agent) ListServices(ctx context.Context, req *mcpv1.ListServicesRequest
 // registry. It returns a list of ServiceInfo messages with updated observed
 // state. This shared logic is used by both LiveCheck and GetServiceStatus.
 func (a *Agent) liveCheckServices(ctx context.Context) ([]*mcpv1.ServiceInfo, error) {
-	containers, err := a.Runtime.List(ctx)
+	containers, err := a.listAllContainers(ctx)
 	if err != nil {
 		return nil, fmt.Errorf("runtime list: %w", err)
 	}
@@ -84,8 +85,9 @@ func (a *Agent) liveCheckServices(ctx context.Context) ([]*mcpv1.ServiceInfo, er
 		}

 		info := &mcpv1.ServiceInfo{
-			Name:   svc.Name,
-			Active: svc.Active,
+			Name:    svc.Name,
+			Active:  svc.Active,
+			Comment: svc.Comment,
 		}

 		for _, comp := range components {
--- a/internal/agent/status_test.go
+++ b/internal/agent/status_test.go
@@ -23,7 +23,7 @@ func TestListServices(t *testing.T) {
 	}

 	// Add a service with components.
-	if err := registry.CreateService(a.DB, "metacrypt", true); err != nil {
+	if err := registry.CreateService(a.DB, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -71,7 +71,7 @@ func TestLiveCheck(t *testing.T) {
 	ctx := context.Background()

 	// Set up registry with one service and one component.
-	if err := registry.CreateService(a.DB, "metacrypt", true); err != nil {
+	if err := registry.CreateService(a.DB, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -146,7 +146,7 @@ func TestGetServiceStatus_DriftDetection(t *testing.T) {
 	a := newTestAgent(t, rt)
 	ctx := context.Background()

-	if err := registry.CreateService(a.DB, "metacrypt", true); err != nil {
+	if err := registry.CreateService(a.DB, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(a.DB, &registry.Component{
@@ -207,7 +207,7 @@ func TestGetServiceStatus_FilterByName(t *testing.T) {
 	ctx := context.Background()

 	for _, svc := range []string{"metacrypt", "mcr"} {
-		if err := registry.CreateService(a.DB, svc, true); err != nil {
+		if err := registry.CreateService(a.DB, svc, true, ""); err != nil {
 			t.Fatalf("create service %q: %v", svc, err)
 		}
 		if err := registry.CreateComponent(a.DB, &registry.Component{
--- a/internal/agent/sync.go
+++ b/internal/agent/sync.go
@@ -60,15 +60,20 @@ func (a *Agent) syncService(_ context.Context, spec *mcpv1.ServiceSpec) (*mcpv1.
 	existing, err := registry.GetService(a.DB, spec.GetName())
 	if err != nil {
 		// Service does not exist; create it.
-		if err := registry.CreateService(a.DB, spec.GetName(), spec.GetActive()); err != nil {
+		if err := registry.CreateService(a.DB, spec.GetName(), spec.GetActive(), spec.GetComment()); err != nil {
 			return nil, nil, status.Errorf(codes.Internal, "create service %q: %v", spec.GetName(), err)
 		}
 		changes = append(changes, "created service")
-	} else if existing.Active != spec.GetActive() {
-		if err := registry.UpdateServiceActive(a.DB, spec.GetName(), spec.GetActive()); err != nil {
+	} else if existing.Active != spec.GetActive() || existing.Comment != spec.GetComment() {
+		if err := registry.UpdateServiceActive(a.DB, spec.GetName(), spec.GetActive(), spec.GetComment()); err != nil {
 			return nil, nil, status.Errorf(codes.Internal, "update service %q: %v", spec.GetName(), err)
 		}
-		changes = append(changes, fmt.Sprintf("active: %v -> %v", existing.Active, spec.GetActive()))
+		if existing.Active != spec.GetActive() {
+			changes = append(changes, fmt.Sprintf("active: %v -> %v", existing.Active, spec.GetActive()))
+		}
+		if existing.Comment != spec.GetComment() {
+			changes = append(changes, fmt.Sprintf("comment: %q", spec.GetComment()))
+		}
 	}

 	// Create or update each component.
@@ -107,7 +112,7 @@ func (a *Agent) syncService(_ context.Context, spec *mcpv1.ServiceSpec) (*mcpv1.
 // reconcileUntracked lists all containers from the runtime and adds any that
 // are not already tracked in the registry with desired_state "ignore".
 func (a *Agent) reconcileUntracked(ctx context.Context, known map[string]bool) error {
-	containers, err := a.Runtime.List(ctx)
+	containers, err := a.listAllContainers(ctx)
 	if err != nil {
 		return fmt.Errorf("list containers: %w", err)
 	}
@@ -127,7 +132,7 @@ func (a *Agent) reconcileUntracked(ctx context.Context, known map[string]bool) e
 		}

 		if _, err := registry.GetService(a.DB, service); err != nil {
-			if err := registry.CreateService(a.DB, service, true); err != nil {
+			if err := registry.CreateService(a.DB, service, true, ""); err != nil {
 				a.Logger.Info("reconcile: create service failed", "service", service, "error", err)
 				continue
 			}
--- a/internal/agent/undeploy.go
+++ b/internal/agent/undeploy.go
@@ -36,7 +36,7 @@ func (a *Agent) UndeployService(ctx context.Context, req *mcpv1.UndeployServiceR
 	}

 	// Mark the service as inactive.
-	if err := registry.UpdateServiceActive(a.DB, serviceName, false); err != nil {
+	if err := registry.UpdateServiceActive(a.DB, serviceName, false, ""); err != nil {
 		a.Logger.Warn("failed to mark service inactive", "service", serviceName, "err", err)
 	}

@@ -73,10 +73,11 @@ func (a *Agent) undeployComponent(ctx context.Context, serviceName string, c *re
 	}

 	// 4. Stop and remove the container.
-	if err := a.Runtime.Stop(ctx, containerName); err != nil {
+	rt := a.runtimeFor(c.Runtime)
+	if err := rt.Stop(ctx, containerName); err != nil {
 		a.Logger.Info("stop container (ignored)", "container", containerName, "error", err)
 	}
-	if err := a.Runtime.Remove(ctx, containerName); err != nil {
+	if err := rt.Remove(ctx, containerName); err != nil {
 		a.Logger.Info("remove container (ignored)", "container", containerName, "error", err)
 	}

--- a/internal/config/agent.go
+++ b/internal/config/agent.go
@@ -10,17 +10,17 @@ import (

 // AgentConfig is the configuration for the mcp-agent daemon.
 type AgentConfig struct {
-	Server    ServerConfig    `toml:"server"`
-	Database  DatabaseConfig  `toml:"database"`
-	MCIAS     MCIASConfig     `toml:"mcias"`
-	Agent     AgentSettings   `toml:"agent"`
-	MCProxy   MCProxyConfig   `toml:"mcproxy"`
-	Metacrypt MetacryptConfig `toml:"metacrypt"`
-	MCNS      MCNSConfig      `toml:"mcns"`
-	Monitor   MonitorConfig   `toml:"monitor"`
-	Log       LogConfig       `toml:"log"`
-	Boot   BootConfig        `toml:"boot"`
-	Master AgentMasterConfig `toml:"master"`
+	Server    ServerConfig      `toml:"server"`
+	Database  DatabaseConfig    `toml:"database"`
+	MCIAS     MCIASConfig       `toml:"mcias"`
+	Agent     AgentSettings     `toml:"agent"`
+	MCProxy   MCProxyConfig     `toml:"mcproxy"`
+	Metacrypt MetacryptConfig   `toml:"metacrypt"`
+	MCNS      MCNSConfig        `toml:"mcns"`
+	Monitor   MonitorConfig     `toml:"monitor"`
+	Log       LogConfig         `toml:"log"`
+	Boot      BootConfig        `toml:"boot"`
+	Master    AgentMasterConfig `toml:"master"`
 }

 // AgentMasterConfig holds the optional master connection settings.
--- a/internal/master/status.go
+++ b/internal/master/status.go
@@ -117,7 +117,7 @@ func (m *Master) ListNodes(_ context.Context, _ *mcpv1.ListNodesRequest) (*mcpv1
 			Address:    n.Address,
 			Arch:       n.Arch,
 			Status:     n.Status,
-			Containers: int32(n.Containers), //nolint:gosec // small number
+			Containers: int32(n.Containers),   //nolint:gosec // small number
 			Services:   int32(counts[n.Name]), //nolint:gosec // small number
 		}
 		if n.LastHeartbeat != nil {
--- a/internal/monitor/monitor_test.go
+++ b/internal/monitor/monitor_test.go
@@ -68,7 +68,7 @@ func TestAlerterDriftDetection(t *testing.T) {
 	al := NewAlerter(cfg, "test-node", db, logger)

 	// Set up a service and component so CountEvents works.
-	if err := registry.CreateService(db, "metacrypt", true); err != nil {
+	if err := registry.CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(db, &registry.Component{
@@ -112,7 +112,7 @@ func TestAlerterCooldownSuppression(t *testing.T) {

 	al := NewAlerter(cfg, "test-node", db, logger)

-	if err := registry.CreateService(db, "metacrypt", true); err != nil {
+	if err := registry.CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(db, &registry.Component{
@@ -148,7 +148,7 @@ func TestAlerterFlapDetection(t *testing.T) {

 	al := NewAlerter(cfg, "test-node", db, logger)

-	if err := registry.CreateService(db, "metacrypt", true); err != nil {
+	if err := registry.CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(db, &registry.Component{
@@ -179,7 +179,7 @@ func TestMonitorTickStateChange(t *testing.T) {
 	logger := testLogger()
 	cfg := testMonitorConfig()

-	if err := registry.CreateService(db, "metacrypt", true); err != nil {
+	if err := registry.CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(db, &registry.Component{
@@ -251,7 +251,7 @@ func TestMonitorNoChangeNoEvent(t *testing.T) {
 	logger := testLogger()
 	cfg := testMonitorConfig()

-	if err := registry.CreateService(db, "metacrypt", true); err != nil {
+	if err := registry.CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}
 	if err := registry.CreateComponent(db, &registry.Component{
--- a/internal/registry/components.go
+++ b/internal/registry/components.go
@@ -30,10 +30,22 @@ type Component struct {
 	Volumes       []string
 	Cmd           []string
 	Routes        []Route
+	Runtime       string // "container" (default) or "unikernel"
+	MemoryMB      int    // unikernel guest memory in MB
+	VCPUs         int    // unikernel guest vCPUs
 	CreatedAt     time.Time
 	UpdatedAt     time.Time
 }

+// defaultRuntime normalizes an empty runtime to "container" so the
+// components.runtime column is never empty.
+func defaultRuntime(r string) string {
+	if r == "" {
+		return "container"
+	}
+	return r
+}
+
 // CreateComponent creates a new component in the registry.
 func CreateComponent(db *sql.DB, c *Component) error {
 	tx, err := db.Begin()
@@ -43,10 +55,10 @@ func CreateComponent(db *sql.DB, c *Component) error {
 	defer tx.Rollback() //nolint:errcheck

 	_, err = tx.Exec(`
-		INSERT INTO components (name, service, image, network, user_spec, restart, desired_state, observed_state, version)
-		VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)`,
+		INSERT INTO components (name, service, image, network, user_spec, restart, desired_state, observed_state, version, runtime, memory_mb, vcpus)
+		VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
 		c.Name, c.Service, c.Image, c.Network, c.UserSpec, c.Restart,
-		c.DesiredState, c.ObservedState, c.Version,
+		c.DesiredState, c.ObservedState, c.Version, defaultRuntime(c.Runtime), c.MemoryMB, c.VCPUs,
 	)
 	if err != nil {
 		return fmt.Errorf("insert component %q/%q: %w", c.Service, c.Name, err)
@@ -74,11 +86,11 @@ func GetComponent(db *sql.DB, service, name string) (*Component, error) {
 	var createdAt, updatedAt string
 	err := db.QueryRow(`
 		SELECT name, service, image, network, user_spec, restart,
-		       desired_state, observed_state, version, created_at, updated_at
+		       desired_state, observed_state, version, runtime, memory_mb, vcpus, created_at, updated_at
 		FROM components WHERE service = ? AND name = ?`,
 		service, name,
 	).Scan(&c.Name, &c.Service, &c.Image, &c.Network, &c.UserSpec, &c.Restart,
-		&c.DesiredState, &c.ObservedState, &c.Version, &createdAt, &updatedAt)
+		&c.DesiredState, &c.ObservedState, &c.Version, &c.Runtime, &c.MemoryMB, &c.VCPUs, &createdAt, &updatedAt)
 	if err != nil {
 		return nil, fmt.Errorf("get component %q/%q: %w", service, name, err)
 	}
@@ -109,7 +121,7 @@ func GetComponent(db *sql.DB, service, name string) (*Component, error) {
 func ListComponents(db *sql.DB, service string) ([]Component, error) {
 	rows, err := db.Query(`
 		SELECT name, service, image, network, user_spec, restart,
-		       desired_state, observed_state, version, created_at, updated_at
+		       desired_state, observed_state, version, runtime, memory_mb, vcpus, created_at, updated_at
 		FROM components WHERE service = ? ORDER BY name`,
 		service,
 	)
@@ -123,7 +135,7 @@ func ListComponents(db *sql.DB, service string) ([]Component, error) {
 		var c Component
 		var createdAt, updatedAt string
 		if err := rows.Scan(&c.Name, &c.Service, &c.Image, &c.Network, &c.UserSpec, &c.Restart,
-			&c.DesiredState, &c.ObservedState, &c.Version, &createdAt, &updatedAt); err != nil {
+			&c.DesiredState, &c.ObservedState, &c.Version, &c.Runtime, &c.MemoryMB, &c.VCPUs, &createdAt, &updatedAt); err != nil {
 			return nil, fmt.Errorf("scan component: %w", err)
 		}
 		c.CreatedAt, _ = time.Parse("2006-01-02 15:04:05", createdAt)
@@ -169,9 +181,11 @@ func UpdateComponentSpec(db *sql.DB, c *Component) error {

 	_, err = tx.Exec(`
 		UPDATE components
-		SET image = ?, network = ?, user_spec = ?, restart = ?, version = ?, updated_at = datetime('now')
+		SET image = ?, network = ?, user_spec = ?, restart = ?, version = ?,
+		    runtime = ?, memory_mb = ?, vcpus = ?, updated_at = datetime('now')
 		WHERE service = ? AND name = ?`,
-		c.Image, c.Network, c.UserSpec, c.Restart, c.Version, c.Service, c.Name,
+		c.Image, c.Network, c.UserSpec, c.Restart, c.Version,
+		defaultRuntime(c.Runtime), c.MemoryMB, c.VCPUs, c.Service, c.Name,
 	)
 	if err != nil {
 		return fmt.Errorf("update component %q/%q: %w", c.Service, c.Name, err)
--- a/internal/registry/db.go
+++ b/internal/registry/db.go
@@ -156,4 +156,9 @@ var migrations = []string{
 		created_at       TEXT NOT NULL DEFAULT (datetime('now')),
 		updated_at       TEXT NOT NULL DEFAULT (datetime('now'))
 	);`,
+
+	// Migration 5: unikernel runtime support (per-component runtime + VM resources)
+	`ALTER TABLE components ADD COLUMN runtime TEXT NOT NULL DEFAULT 'container';
+	 ALTER TABLE components ADD COLUMN memory_mb INTEGER NOT NULL DEFAULT 0;
+	 ALTER TABLE components ADD COLUMN vcpus INTEGER NOT NULL DEFAULT 0;`,
 }
--- a/internal/registry/registry_test.go
+++ b/internal/registry/registry_test.go
@@ -37,7 +37,7 @@ func TestServiceCRUD(t *testing.T) {
 	db := openTestDB(t)

 	// Create
-	if err := CreateService(db, "metacrypt", true); err != nil {
+	if err := CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create: %v", err)
 	}

@@ -60,7 +60,7 @@ func TestServiceCRUD(t *testing.T) {
 	}

 	// Update active
-	if err := UpdateServiceActive(db, "metacrypt", false); err != nil {
+	if err := UpdateServiceActive(db, "metacrypt", false, ""); err != nil {
 		t.Fatalf("update: %v", err)
 	}
 	s, _ = GetService(db, "metacrypt")
@@ -80,17 +80,17 @@ func TestServiceCRUD(t *testing.T) {

 func TestServiceDuplicateName(t *testing.T) {
 	db := openTestDB(t)
-	if err := CreateService(db, "metacrypt", true); err != nil {
+	if err := CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("first create: %v", err)
 	}
-	if err := CreateService(db, "metacrypt", true); err == nil {
+	if err := CreateService(db, "metacrypt", true, ""); err == nil {
 		t.Fatal("expected error on duplicate name")
 	}
 }

 func TestComponentCRUD(t *testing.T) {
 	db := openTestDB(t)
-	if err := CreateService(db, "metacrypt", true); err != nil {
+	if err := CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}

@@ -198,7 +198,7 @@ func TestComponentCRUD(t *testing.T) {

 func TestComponentCompositePK(t *testing.T) {
 	db := openTestDB(t)
-	if err := CreateService(db, "metacrypt", true); err != nil {
+	if err := CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}

@@ -213,7 +213,7 @@ func TestComponentCompositePK(t *testing.T) {

 func TestCascadeDelete(t *testing.T) {
 	db := openTestDB(t)
-	if err := CreateService(db, "metacrypt", true); err != nil {
+	if err := CreateService(db, "metacrypt", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}

@@ -239,7 +239,7 @@ func TestCascadeDelete(t *testing.T) {

 func TestComponentRoutes(t *testing.T) {
 	db := openTestDB(t)
-	if err := CreateService(db, "svc", true); err != nil {
+	if err := CreateService(db, "svc", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}

@@ -298,7 +298,7 @@ func TestComponentRoutes(t *testing.T) {

 func TestRouteHostPort(t *testing.T) {
 	db := openTestDB(t)
-	if err := CreateService(db, "svc", true); err != nil {
+	if err := CreateService(db, "svc", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}

@@ -363,7 +363,7 @@ func TestRouteHostPort(t *testing.T) {

 func TestRouteCascadeDelete(t *testing.T) {
 	db := openTestDB(t)
-	if err := CreateService(db, "svc", true); err != nil {
+	if err := CreateService(db, "svc", true, ""); err != nil {
 		t.Fatalf("create service: %v", err)
 	}

--- a/internal/registry/services.go
+++ b/internal/registry/services.go
@@ -10,15 +10,16 @@ import (
 type Service struct {
 	Name      string
 	Active    bool
+	Comment   string
 	CreatedAt time.Time
 	UpdatedAt time.Time
 }

 // CreateService creates a new service in the registry.
-func CreateService(db *sql.DB, name string, active bool) error {
+func CreateService(db *sql.DB, name string, active bool, comment string) error {
 	_, err := db.Exec(
-		"INSERT INTO services (name, active) VALUES (?, ?)",
-		name, active,
+		"INSERT INTO services (name, active, comment) VALUES (?, ?, ?)",
+		name, active, comment,
 	)
 	if err != nil {
 		return fmt.Errorf("create service %q: %w", name, err)
@@ -31,9 +32,9 @@ func GetService(db *sql.DB, name string) (*Service, error) {
 	s := &Service{}
 	var createdAt, updatedAt string
 	err := db.QueryRow(
-		"SELECT name, active, created_at, updated_at FROM services WHERE name = ?",
+		"SELECT name, active, comment, created_at, updated_at FROM services WHERE name = ?",
 		name,
-	).Scan(&s.Name, &s.Active, &createdAt, &updatedAt)
+	).Scan(&s.Name, &s.Active, &s.Comment, &createdAt, &updatedAt)
 	if err != nil {
 		return nil, fmt.Errorf("get service %q: %w", name, err)
 	}
@@ -44,7 +45,7 @@ func GetService(db *sql.DB, name string) (*Service, error) {

 // ListServices returns all services.
 func ListServices(db *sql.DB) ([]Service, error) {
-	rows, err := db.Query("SELECT name, active, created_at, updated_at FROM services ORDER BY name")
+	rows, err := db.Query("SELECT name, active, comment, created_at, updated_at FROM services ORDER BY name")
 	if err != nil {
 		return nil, fmt.Errorf("list services: %w", err)
 	}
@@ -54,7 +55,7 @@ func ListServices(db *sql.DB) ([]Service, error) {
 	for rows.Next() {
 		var s Service
 		var createdAt, updatedAt string
-		if err := rows.Scan(&s.Name, &s.Active, &createdAt, &updatedAt); err != nil {
+		if err := rows.Scan(&s.Name, &s.Active, &s.Comment, &createdAt, &updatedAt); err != nil {
 			return nil, fmt.Errorf("scan service: %w", err)
 		}
 		s.CreatedAt, _ = time.Parse("2006-01-02 15:04:05", createdAt)
@@ -64,11 +65,15 @@ func ListServices(db *sql.DB) ([]Service, error) {
 	return services, rows.Err()
 }

-// UpdateServiceActive updates a service's active flag.
-func UpdateServiceActive(db *sql.DB, name string, active bool) error {
-	res, err := db.Exec(
-		"UPDATE services SET active = ?, updated_at = datetime('now') WHERE name = ?",
-		active, name,
+// UpdateServiceActive updates a service's active flag and comment. If comment
+// is empty, the existing comment is preserved.
+func UpdateServiceActive(db *sql.DB, name string, active bool, comment string) error {
+	res, err := db.Exec(`
+		UPDATE services SET active = ?,
+			comment = CASE WHEN ? = '' THEN comment ELSE ? END,
+			updated_at = datetime('now')
+		WHERE name = ?`,
+		active, comment, comment, name,
 	)
 	if err != nil {
 		return fmt.Errorf("update service %q: %w", name, err)
--- a/internal/runtime/qemu.go
+++ b/internal/runtime/qemu.go
@@ -0,0 +1,558 @@
+package runtime
+
+import (
+	"context"
+	"crypto/sha256"
+	"encoding/hex"
+	"encoding/json"
+	"fmt"
+	"net"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"syscall"
+	"time"
+)
+
+// dialUnix connects to a unix-domain socket with a timeout.
+func dialUnix(path string, timeout time.Duration) (net.Conn, error) {
+	return net.DialTimeout("unix", path, timeout)
+}
+
+// QEMU implements the Runtime interface by running services as Nanos
+// unikernel virtual machines under QEMU/KVM instead of containers.
+//
+// Each service component becomes a single-process VM with its own kernel.
+// The lifecycle maps onto the same Runtime interface as Podman so the agent
+// can treat unikernels and containers uniformly:
+//
+//	Pull    -> pull the OCI image, extract the ELF binary, cache it
+//	Run     -> `ops build` the binary into a Nanos image, boot it under QEMU
+//	Stop    -> graceful QMP powerdown, then SIGTERM/SIGKILL the QEMU process
+//	Remove  -> stop and delete the VM state directory
+//	Inspect -> read persisted metadata + check process liveness
+//	List    -> enumerate VM state directories
+//	Logs    -> stream the serial console log file
+//
+// Phase 1 uses QEMU user-mode networking with host port forwards, which is
+// functionally equivalent to rootless podman's localhost port mappings:
+// mc-proxy routes to 127.0.0.1:<hostport> exactly as it does for containers.
+// Isolated bridge networking is a later phase.
+type QEMU struct {
+	// ImageDir holds built Nanos images and extracted binaries.
+	// Default: /srv/mcp/images
+	ImageDir string
+	// StateDir holds per-VM runtime state (pidfile, QMP socket, console log,
+	// metadata). Default: /srv/mcp/vm
+	StateDir string
+	// OpsPath is the path to the `ops` Nanos toolchain binary. Default: "ops".
+	OpsPath string
+	// QemuPath is the path to qemu-system-x86_64. Default: "qemu-system-x86_64".
+	QemuPath string
+	// Memory is the default guest memory in MB when a spec does not set one.
+	Memory int
+	// HomeDir is set as $HOME for `ops` so it uses a stable ~/.ops directory.
+	HomeDir string
+}
+
+func (q *QEMU) imageDir() string {
+	if q.ImageDir != "" {
+		return q.ImageDir
+	}
+	return "/srv/mcp/images"
+}
+
+func (q *QEMU) stateDir() string {
+	if q.StateDir != "" {
+		return q.StateDir
+	}
+	return "/srv/mcp/vm"
+}
+
+func (q *QEMU) opsPath() string {
+	if q.OpsPath != "" {
+		return q.OpsPath
+	}
+	return "ops"
+}
+
+func (q *QEMU) qemuPath() string {
+	if q.QemuPath != "" {
+		return q.QemuPath
+	}
+	return "qemu-system-x86_64"
+}
+
+func (q *QEMU) memory() int {
+	if q.Memory > 0 {
+		return q.Memory
+	}
+	return 256
+}
+
+// opsEnv returns the environment for invoking `ops`, pinning $HOME so its
+// cache and image directory are stable across invocations.
+func (q *QEMU) opsEnv() []string {
+	env := os.Environ()
+	if q.HomeDir != "" {
+		env = append(env, "HOME="+q.HomeDir)
+	}
+	return env
+}
+
+// sanitizeImage turns an image reference into a filesystem-safe stem.
+//
+//	"mcr.example:8443/mcdoc:v0.1.0" -> "mcr.example_8443_mcdoc_v0.1.0"
+func sanitizeImage(image string) string {
+	r := strings.NewReplacer("/", "_", ":", "_")
+	return r.Replace(image)
+}
+
+// binaryName derives the in-image ELF binary name from an image reference by
+// taking the repository basename. "host:8443/mcdoc:v0.1.0" -> "mcdoc".
+func binaryName(image string) string {
+	name := image
+	if i := strings.LastIndex(name, "/"); i >= 0 {
+		name = name[i+1:]
+	}
+	if i := strings.Index(name, ":"); i >= 0 {
+		name = name[:i]
+	}
+	return name
+}
+
+func (q *QEMU) binPath(image string) string {
+	return filepath.Join(q.imageDir(), sanitizeImage(image)+".bin")
+}
+
+func (q *QEMU) imgPath(name string) string {
+	return filepath.Join(q.imageDir(), name+".img")
+}
+
+func (q *QEMU) vmDir(name string) string {
+	return filepath.Join(q.stateDir(), name)
+}
+
+// vmMeta is the persisted per-VM metadata written at Run time so that
+// Inspect/List can report accurate information after an agent restart.
+type vmMeta struct {
+	Name      string    `json:"name"`
+	Image     string    `json:"image"`
+	User      string    `json:"user"`
+	Restart   string    `json:"restart"`
+	Ports     []string  `json:"ports"`
+	Volumes   []string  `json:"volumes"`
+	Cmd       []string  `json:"cmd"`
+	MemoryMB  int       `json:"memory_mb"`
+	VCPUs     int       `json:"vcpus"`
+	ImageHash string    `json:"image_hash"`
+	Started   time.Time `json:"started"`
+}
+
+// Pull pulls the OCI image and extracts its ELF binary into the image cache.
+// The binary is the input to `ops build`; the Nanos image itself is built at
+// Run time so the service's command arguments can be baked in.
+func (q *QEMU) Pull(ctx context.Context, image string) error {
+	if err := os.MkdirAll(q.imageDir(), 0o750); err != nil {
+		return fmt.Errorf("create image dir: %w", err)
+	}
+
+	// Pull the OCI image via podman (reuses the agent's registry auth).
+	if out, err := exec.CommandContext(ctx, "podman", "pull", image).CombinedOutput(); err != nil { //nolint:gosec // args built programmatically
+		return fmt.Errorf("podman pull %q: %w: %s", image, err, out)
+	}
+
+	// Create (do not start) a container to copy the binary out of.
+	tmp := "ukextract-" + sanitizeImage(image)
+	_ = exec.CommandContext(ctx, "podman", "rm", "-f", tmp).Run()                                                    //nolint:gosec
+	if out, err := exec.CommandContext(ctx, "podman", "create", "--name", tmp, image).CombinedOutput(); err != nil { //nolint:gosec
+		return fmt.Errorf("podman create %q: %w: %s", image, err, out)
+	}
+	defer func() { _ = exec.Command("podman", "rm", "-f", tmp).Run() }() //nolint:gosec
+
+	bin := binaryName(image)
+	src := tmp + ":/usr/local/bin/" + bin
+	dst := q.binPath(image)
+	if out, err := exec.CommandContext(ctx, "podman", "cp", src, dst).CombinedOutput(); err != nil { //nolint:gosec
+		return fmt.Errorf("extract binary %q from %q: %w: %s", bin, image, err, out)
+	}
+	if err := os.Chmod(dst, 0o755); err != nil { //nolint:gosec // unikernel ELF must be executable
+		return fmt.Errorf("chmod extracted binary: %w", err)
+	}
+	return nil
+}
+
+// opsConfig is the subset of the `ops` build configuration we generate.
+type opsConfig struct {
+	Args      []string          `json:"Args,omitempty"`
+	Env       map[string]string `json:"Env,omitempty"`
+	RunConfig opsRunConfig      `json:"RunConfig"`
+}
+
+type opsRunConfig struct {
+	Ports    []string `json:"Ports,omitempty"`
+	Memory   string   `json:"Memory,omitempty"`
+	CPUs     int      `json:"CPUs,omitempty"`
+	Klibs    []string `json:"Klibs,omitempty"`
+	Mounts   any      `json:"Mounts,omitempty"`
+	NoTrace  []string `json:"NoTrace,omitempty"`
+	GDBPort  int      `json:"GDBPort,omitempty"`
+	Nanos    string   `json:"Nanos,omitempty"`
+	Hostname string   `json:"Hostname,omitempty"`
+}
+
+// guestPorts extracts the guest (container) port from each spec port mapping.
+// Accepts "host:container", "ip:host:container", or a bare "port".
+func guestPorts(ports []string) []string {
+	var gp []string
+	for _, p := range ports {
+		parts := strings.Split(p, ":")
+		gp = append(gp, parts[len(parts)-1])
+	}
+	return gp
+}
+
+// hostForward builds the QEMU hostfwd value for a spec port mapping.
+// "ip:host:container" -> "tcp:ip:host-:container"
+// "host:container"    -> "tcp:127.0.0.1:host-:container"
+func hostForward(p string) string {
+	parts := strings.Split(p, ":")
+	switch len(parts) {
+	case 3:
+		return fmt.Sprintf("tcp:%s:%s-:%s", parts[0], parts[1], parts[2])
+	case 2:
+		return fmt.Sprintf("tcp:127.0.0.1:%s-:%s", parts[0], parts[1])
+	default:
+		return fmt.Sprintf("tcp:127.0.0.1:%s-:%s", parts[0], parts[0])
+	}
+}
+
+// Run builds the Nanos image (if needed) and boots it under QEMU/KVM.
+func (q *QEMU) Run(ctx context.Context, spec ContainerSpec) error {
+	if err := os.MkdirAll(q.vmDir(spec.Name), 0o750); err != nil {
+		return fmt.Errorf("create vm state dir: %w", err)
+	}
+
+	mem := spec.MemoryMB
+	if mem <= 0 {
+		mem = q.memory()
+	}
+	cpus := spec.VCPUs
+	if cpus <= 0 {
+		cpus = 1
+	}
+
+	// Build the Nanos image from the extracted binary, baking in command args.
+	cfg := opsConfig{
+		Args: spec.Cmd,
+		RunConfig: opsRunConfig{
+			Ports:  guestPorts(spec.Ports),
+			Memory: strconv.Itoa(mem) + "m",
+			CPUs:   cpus,
+		},
+	}
+	if len(spec.Env) > 0 {
+		cfg.Env = map[string]string{}
+		for _, e := range spec.Env {
+			if i := strings.Index(e, "="); i >= 0 {
+				cfg.Env[e[:i]] = e[i+1:]
+			}
+		}
+	}
+	cfgPath := filepath.Join(q.vmDir(spec.Name), "ops.json")
+	cfgBytes, err := json.MarshalIndent(cfg, "", "  ")
+	if err != nil {
+		return fmt.Errorf("marshal ops config: %w", err)
+	}
+	if err := os.WriteFile(cfgPath, cfgBytes, 0o640); err != nil { //nolint:gosec // mcp-group-readable config
+		return fmt.Errorf("write ops config: %w", err)
+	}
+
+	img := q.imgPath(spec.Name)
+	bin := q.binPath(spec.Image)
+	if _, err := os.Stat(bin); err != nil {
+		return fmt.Errorf("binary not found for %q (Pull first): %w", spec.Image, err)
+	}
+	build := exec.CommandContext(ctx, q.opsPath(), "build", bin, "-c", cfgPath, "-i", spec.Name) //nolint:gosec
+	build.Env = q.opsEnv()
+	if out, err := build.CombinedOutput(); err != nil {
+		return fmt.Errorf("ops build %q: %w: %s", spec.Name, err, out)
+	}
+	// ops writes to ~/.ops/images/<name>.img; move it into our image dir.
+	opsImg := filepath.Join(q.opsImagesDir(), spec.Name+".img")
+	if _, err := os.Stat(opsImg); err == nil {
+		if err := os.Rename(opsImg, img); err != nil {
+			// Cross-device fallback: copy.
+			if cpErr := copyFile(opsImg, img); cpErr != nil {
+				return fmt.Errorf("relocate built image: %w", err)
+			}
+			_ = os.Remove(opsImg)
+		}
+	}
+
+	hash, _ := fileSHA256(img)
+
+	// Assemble QEMU invocation: KVM-accelerated, headless, serial console to a
+	// file, QMP control socket, virtio disk + NIC with user-mode port forwards.
+	netdev := "user,id=n0"
+	for _, p := range spec.Ports {
+		netdev += ",hostfwd=" + hostForward(p)
+	}
+	args := []string{
+		"-enable-kvm",
+		"-m", strconv.Itoa(mem),
+		"-smp", strconv.Itoa(cpus),
+		"-display", "none",
+		"-no-reboot",
+		"-daemonize",
+		"-pidfile", filepath.Join(q.vmDir(spec.Name), "qemu.pid"),
+		"-serial", "file:" + filepath.Join(q.vmDir(spec.Name), "console.log"),
+		"-qmp", "unix:" + filepath.Join(q.vmDir(spec.Name), "qmp.sock") + ",server,nowait",
+		"-drive", "file=" + img + ",format=raw,if=virtio",
+		"-device", "virtio-net-pci,netdev=n0",
+		"-netdev", netdev,
+	}
+	// 9p passthrough for host /srv/<service> volumes (best-effort; Nanos must
+	// support the 9p client for the guest to mount it).
+	for i, v := range spec.Volumes {
+		parts := strings.SplitN(v, ":", 2)
+		host := parts[0]
+		tag := fmt.Sprintf("srv%d", i)
+		args = append(args,
+			"-virtfs", fmt.Sprintf("local,path=%s,mount_tag=%s,security_model=none,id=%s", host, tag, tag),
+		)
+	}
+
+	cmd := exec.CommandContext(ctx, q.qemuPath(), args...) //nolint:gosec
+	if out, err := cmd.CombinedOutput(); err != nil {
+		return fmt.Errorf("qemu launch %q: %w: %s", spec.Name, err, out)
+	}
+
+	meta := vmMeta{
+		Name:      spec.Name,
+		Image:     spec.Image,
+		User:      spec.User,
+		Restart:   spec.Restart,
+		Ports:     spec.Ports,
+		Volumes:   spec.Volumes,
+		Cmd:       spec.Cmd,
+		MemoryMB:  mem,
+		VCPUs:     cpus,
+		ImageHash: hash,
+		Started:   time.Now().UTC(),
+	}
+	return q.writeMeta(spec.Name, meta)
+}
+
+func (q *QEMU) opsImagesDir() string {
+	home := q.HomeDir
+	if home == "" {
+		home, _ = os.UserHomeDir()
+	}
+	return filepath.Join(home, ".ops", "images")
+}
+
+func (q *QEMU) writeMeta(name string, m vmMeta) error {
+	b, err := json.MarshalIndent(m, "", "  ")
+	if err != nil {
+		return fmt.Errorf("marshal vm meta: %w", err)
+	}
+	return os.WriteFile(filepath.Join(q.vmDir(name), "meta.json"), b, 0o640) //nolint:gosec // mcp-group-readable metadata
+}
+
+func (q *QEMU) readMeta(name string) (vmMeta, error) {
+	var m vmMeta
+	b, err := os.ReadFile(filepath.Join(q.vmDir(name), "meta.json"))
+	if err != nil {
+		return m, err
+	}
+	return m, json.Unmarshal(b, &m)
+}
+
+// pidOf returns the running QEMU pid for a VM, or 0 if not running.
+func (q *QEMU) pidOf(name string) int {
+	b, err := os.ReadFile(filepath.Join(q.vmDir(name), "qemu.pid"))
+	if err != nil {
+		return 0
+	}
+	pid, err := strconv.Atoi(strings.TrimSpace(string(b)))
+	if err != nil || pid <= 0 {
+		return 0
+	}
+	if err := syscall.Kill(pid, 0); err != nil {
+		return 0
+	}
+	return pid
+}
+
+// Stop gracefully powers down the VM, falling back to SIGTERM/SIGKILL.
+func (q *QEMU) Stop(ctx context.Context, name string) error {
+	pid := q.pidOf(name)
+	if pid == 0 {
+		return nil
+	}
+	// Try a graceful QMP system_powerdown.
+	_ = q.qmpCommand(name, "system_powerdown")
+	deadline := time.Now().Add(10 * time.Second)
+	for time.Now().Before(deadline) {
+		if q.pidOf(name) == 0 {
+			return nil
+		}
+		time.Sleep(300 * time.Millisecond)
+	}
+	// Escalate.
+	_ = syscall.Kill(pid, syscall.SIGTERM)
+	time.Sleep(2 * time.Second)
+	if q.pidOf(name) != 0 {
+		_ = syscall.Kill(pid, syscall.SIGKILL)
+	}
+	return nil
+}
+
+// qmpCommand sends a single QMP command over the VM's control socket.
+func (q *QEMU) qmpCommand(name, command string) error {
+	sock := filepath.Join(q.vmDir(name), "qmp.sock")
+	conn, err := dialUnix(sock, 3*time.Second)
+	if err != nil {
+		return err
+	}
+	defer func() { _ = conn.Close() }()
+	// QMP handshake: read greeting, send qmp_capabilities, then the command.
+	dec := json.NewDecoder(conn)
+	var greeting map[string]any
+	_ = dec.Decode(&greeting)
+	_, _ = conn.Write([]byte(`{"execute":"qmp_capabilities"}`))
+	var ack map[string]any
+	_ = dec.Decode(&ack)
+	_, err = conn.Write([]byte(`{"execute":"` + command + `"}`))
+	return err
+}
+
+// Remove stops the VM and deletes its state directory.
+func (q *QEMU) Remove(ctx context.Context, name string) error {
+	_ = q.Stop(ctx, name)
+	if pid := q.pidOf(name); pid != 0 {
+		_ = syscall.Kill(pid, syscall.SIGKILL)
+	}
+	return os.RemoveAll(q.vmDir(name))
+}
+
+// Inspect reports the observed state of a VM.
+func (q *QEMU) Inspect(ctx context.Context, name string) (ContainerInfo, error) {
+	m, err := q.readMeta(name)
+	if err != nil {
+		return ContainerInfo{}, fmt.Errorf("qemu inspect %q: %w", name, err)
+	}
+	state := "stopped"
+	if q.pidOf(name) != 0 {
+		state = "running"
+	}
+	return q.infoFromMeta(m, state), nil
+}
+
+func (q *QEMU) infoFromMeta(m vmMeta, state string) ContainerInfo {
+	return ContainerInfo{
+		Name:    m.Name,
+		Image:   m.Image,
+		State:   state,
+		Network: "user",
+		User:    m.User,
+		Restart: m.Restart,
+		Ports:   m.Ports,
+		Volumes: m.Volumes,
+		Cmd:     m.Cmd,
+		Version: ExtractVersion(m.Image),
+		Started: m.Started,
+	}
+}
+
+// List enumerates all VMs known from the state directory.
+func (q *QEMU) List(ctx context.Context) ([]ContainerInfo, error) {
+	entries, err := os.ReadDir(q.stateDir())
+	if err != nil {
+		if os.IsNotExist(err) {
+			return nil, nil
+		}
+		return nil, fmt.Errorf("read vm state dir: %w", err)
+	}
+	var infos []ContainerInfo
+	for _, e := range entries {
+		if !e.IsDir() {
+			continue
+		}
+		m, err := q.readMeta(e.Name())
+		if err != nil {
+			continue
+		}
+		state := "stopped"
+		if q.pidOf(e.Name()) != 0 {
+			state = "running"
+		}
+		infos = append(infos, q.infoFromMeta(m, state))
+	}
+	return infos, nil
+}
+
+// Logs streams the VM's serial console log.
+func (q *QEMU) Logs(ctx context.Context, name string, tail int, follow, timestamps bool, since string) *exec.Cmd {
+	console := filepath.Join(q.vmDir(name), "console.log")
+	args := []string{}
+	if follow {
+		args = append(args, "-f")
+	}
+	if tail > 0 {
+		args = append(args, "-n", strconv.Itoa(tail))
+	} else {
+		args = append(args, "-n", "+1")
+	}
+	args = append(args, console)
+	return exec.CommandContext(ctx, "tail", args...) //nolint:gosec
+}
+
+// Build builds a Nanos image from a context directory's binary. Used by the
+// `mcp build --unikernel` path. Not the primary deploy path.
+func (q *QEMU) Build(ctx context.Context, image, contextDir, dockerfile string) error {
+	return fmt.Errorf("qemu build: not implemented; build OCI image then Pull")
+}
+
+// Push is not implemented for the QEMU runtime.
+func (q *QEMU) Push(ctx context.Context, image string) error {
+	return fmt.Errorf("qemu push: not implemented")
+}
+
+// ImageExists reports whether the extracted binary for the image is cached.
+func (q *QEMU) ImageExists(ctx context.Context, image string) (bool, error) {
+	if _, err := os.Stat(q.binPath(image)); err == nil {
+		return true, nil
+	}
+	return false, nil
+}
+
+// Login delegates registry auth to podman (shared credential store).
+func (q *QEMU) Login(ctx context.Context, registry, username, token string) error {
+	cmd := exec.CommandContext(ctx, "podman", "login", "--username", username, "--password-stdin", registry) //nolint:gosec
+	cmd.Stdin = strings.NewReader(token)
+	if out, err := cmd.CombinedOutput(); err != nil {
+		return fmt.Errorf("podman login %q: %w: %s", registry, err, out)
+	}
+	return nil
+}
+
+func fileSHA256(path string) (string, error) {
+	b, err := os.ReadFile(path) //nolint:gosec // hashing a known image path
+	if err != nil {
+		return "", err
+	}
+	sum := sha256.Sum256(b)
+	return hex.EncodeToString(sum[:]), nil
+}
+
+func copyFile(src, dst string) error {
+	in, err := os.ReadFile(src) //nolint:gosec // relocating a built image
+	if err != nil {
+		return err
+	}
+	return os.WriteFile(dst, in, 0o640) //nolint:gosec // relocating a built image
+}
--- a/internal/runtime/qemu_test.go
+++ b/internal/runtime/qemu_test.go
@@ -0,0 +1,43 @@
+package runtime
+
+import "testing"
+
+func TestSanitizeImage(t *testing.T) {
+	got := sanitizeImage("mcr.example:8443/mcdoc:v0.1.0")
+	if got != "mcr.example_8443_mcdoc_v0.1.0" {
+		t.Errorf("sanitizeImage = %q", got)
+	}
+}
+
+func TestBinaryName(t *testing.T) {
+	cases := map[string]string{
+		"host:8443/mcdoc:v0.1.0": "mcdoc",
+		"mcdoc:v1":               "mcdoc",
+		"reg/uktest":             "uktest",
+	}
+	for in, want := range cases {
+		if got := binaryName(in); got != want {
+			t.Errorf("binaryName(%q) = %q, want %q", in, got, want)
+		}
+	}
+}
+
+func TestHostForward(t *testing.T) {
+	cases := map[string]string{
+		"100.88.197.9:18080:8080": "tcp:100.88.197.9:18080-:8080",
+		"18080:8080":              "tcp:127.0.0.1:18080-:8080",
+		"8080":                    "tcp:127.0.0.1:8080-:8080",
+	}
+	for in, want := range cases {
+		if got := hostForward(in); got != want {
+			t.Errorf("hostForward(%q) = %q, want %q", in, got, want)
+		}
+	}
+}
+
+func TestGuestPorts(t *testing.T) {
+	got := guestPorts([]string{"100.88.197.9:18080:8080", "9090"})
+	if len(got) != 2 || got[0] != "8080" || got[1] != "9090" {
+		t.Errorf("guestPorts = %v", got)
+	}
+}
--- a/internal/runtime/runtime.go
+++ b/internal/runtime/runtime.go
@@ -17,6 +17,10 @@ type ContainerSpec struct {
 	Volumes []string // "host:container" volume mounts
 	Cmd     []string // command and arguments
 	Env     []string // environment variables (KEY=VALUE)
+
+	// Unikernel-only fields (ignored by the container runtime).
+	MemoryMB int // guest memory in MB (default 256)
+	VCPUs    int // guest vCPUs (default 1)
 }

 // ContainerInfo describes the observed state of a running or stopped container.
--- a/internal/servicedef/servicedef.go
+++ b/internal/servicedef/servicedef.go
@@ -53,6 +53,9 @@ type ComponentDef struct {
 	Cmd     []string   `toml:"cmd,omitempty"`
 	Routes  []RouteDef `toml:"routes,omitempty"`
 	Env     []string   `toml:"env,omitempty"`
+	Runtime string     `toml:"runtime,omitempty"` // "container" (default) or "unikernel"
+	Memory  int        `toml:"memory,omitempty"`  // unikernel guest memory in MB
+	VCPUs   int        `toml:"vcpus,omitempty"`   // unikernel guest vCPUs
 }

 // Load reads and parses a TOML service definition file. If the active field
@@ -204,15 +207,18 @@ func ToProto(def *ServiceDef) *mcpv1.ServiceSpec {

 	for _, c := range def.Components {
 		cs := &mcpv1.ComponentSpec{
-			Name:    c.Name,
-			Image:   c.Image,
-			Network: c.Network,
-			User:    c.User,
-			Restart: c.Restart,
-			Ports:   c.Ports,
-			Volumes: c.Volumes,
-			Cmd:     c.Cmd,
-			Env:     c.Env,
+			Name:     c.Name,
+			Image:    c.Image,
+			Network:  c.Network,
+			User:     c.User,
+			Restart:  c.Restart,
+			Ports:    c.Ports,
+			Volumes:  c.Volumes,
+			Cmd:      c.Cmd,
+			Env:      c.Env,
+			Runtime:  c.Runtime,
+			MemoryMb: int32(c.Memory), //nolint:gosec // small config value
+			Vcpus:    int32(c.VCPUs),  //nolint:gosec // small config value
 		}
 		for _, r := range c.Routes {
 			cs.Routes = append(cs.Routes, &mcpv1.RouteSpec{