feat(edge): node resource model unification - update phase/SDD docs and edge service

- Move node-resource-model-unification milestone and SDD to archive
- Update PHASE.md for operational-observability-provider-management
- Refactor edge node registration and model queue logic
- Add status provider tests for model queue validation
- Update edge config and Go config structs

agent-roadmap/archive/phase/operational-observability-provider-management/milestones/node-resource-model-unification.md

@@ -13,7 +13,7 @@ Mac dev-runtime에서는 Codex CLI와 MLX vLLM provider를 하나의 `mac-codex-
 
 ## 상태
 
-[검토중]
+[완료]
 
 ## 구현 잠금
 
@@ -51,11 +51,11 @@ Node를 Edge 연결 owner로 두고, 실행 제어는 Node 하위 resource/provi
 
 ## 완료 리뷰
 
-- 상태: 검토중
+- 상태: 통과
 - 요청일: 2026-06-29
-- 완료 근거: `contract-schema`, `node-admission`, `edge-routing-status`, `dev-runtime-profile` 모두 PASS complete log와 `Roadmap Completion`을 남겼고, SDD Acceptance Scenario S01-S07과 Evidence Map을 충족한다.
-- 검토 항목: archive 전환 전 최종 완료 확인
-- 리뷰 코멘트: 남은 차단 항목 없음. 완료 archive 전환은 별도 승인 후 진행한다.
+- 완료 근거: `contract-schema`, `node-admission`, `edge-routing-status`, `dev-runtime-profile` 모두 PASS complete log와 `Roadmap Completion`을 남겼고, SDD Acceptance Scenario S01-S07과 Evidence Map을 충족한다. 종료 전 코드레벨 검토에서 provider catalog zero-capacity status fallback과 stale CLI help 문구를 보정했고 `go test -count=1 ./...`가 통과했다.
+- 검토 항목: 완료 archive 전환
+- 리뷰 코멘트: 남은 차단 항목 없음. 2026-06-29 완료 archive로 이동한다.
 
 ## 범위 제외
 

agent-roadmap/phase/operational-observability-provider-management/PHASE.md

@@ -20,9 +20,9 @@ provider 확장 Phase에서 검증한 Ollama, vLLM, SGLang, Lemonade 같은 추
   - 경로: `agent-roadmap/archive/phase/operational-observability-provider-management/milestones/provider-catalog-device-status.md`
   - 요약: `models[]`를 외부 model key이자 운영 catalog model로 두고, Node 하위 provider와 served model list를 기준으로 Edge가 provider 선택, load-ratio routing, target rewrite, read-only catalog 상태를 소유하는 MVP 계약을 완료했다.
 
-- [검토중] Node Resource Model Unification
-  - 경로: `agent-roadmap/phase/operational-observability-provider-management/milestones/node-resource-model-unification.md`
-  - 요약: Node를 Edge 연결 identity로 두고 CLI, OpenAI-compatible provider, 기타 resource를 같은 Node 아래 나열하며 provider/resource capacity만 concurrency를 소유하도록 runtime 계약과 dev-runtime 구성을 정렬했고, 완료 근거와 archive 전환 검토를 남긴다.
+- [완료] Node Resource Model Unification
+  - 경로: `agent-roadmap/archive/phase/operational-observability-provider-management/milestones/node-resource-model-unification.md`
+  - 요약: Node를 Edge 연결 identity로 두고 CLI, OpenAI-compatible provider, 기타 resource를 같은 Node 아래 나열하며 provider/resource capacity만 concurrency를 소유하도록 runtime 계약과 dev-runtime 구성을 정렬했다.
 
 - [스케치] 사용량, 토큰, 로그 운영 추적 MVP
   - 경로: `agent-roadmap/phase/operational-observability-provider-management/milestones/usage-token-log-ops-mvp.md`

apps/edge/internal/edgecmd/node_register.go

@@ -264,7 +264,7 @@ directly in the default flow.`,
 	c.Flags().StringVar(&regAdapter, "adapter", "cli", "node adapter: ollama or cli")
 	c.Flags().StringVar(&regOllamaBaseURL, "ollama-base-url", "", "ollama base URL")
 	c.Flags().IntVar(&regOllamaContextSize, "ollama-context-size", 2048, "ollama context size")
-	c.Flags().IntVar(&regRuntimeConcurrency, "runtime-concurrency", 1, "runtime concurrency limit")
+	c.Flags().IntVar(&regRuntimeConcurrency, "runtime-concurrency", 1, "legacy runtime concurrency metadata")
 	c.Flags().StringVar(&regTarget, "target", "", "bootstrap target platform (e.g. darwin-arm64, linux-amd64)")
 	c.Flags().StringVar(&regArtifactBaseURL, "artifact-base-url", "", "artifact base URL for bootstrap download")
 

apps/edge/internal/service/model_queue.go

@@ -532,11 +532,6 @@ func (m *modelQueueManager) getSnapshotForNode(nodeID string, rec *edgenode.Node
 
 	var snaps []*iop.ProviderSnapshot
 
-	concurrencyFallback := 1
-	if rec.Runtime.Concurrency > 0 {
-		concurrencyFallback = rec.Runtime.Concurrency
-	}
-
 	// Catalog-first: nodes with a providers[] catalog emit only catalog snapshots.
 	// Adapter snapshots (sections 1-4) are omitted to prevent duplicate metrics
 	// when a node has both adapter instances and provider-pool catalog entries.
@@ -546,9 +541,6 @@ func (m *modelQueueManager) getSnapshotForNode(nodeID string, rec *edgenode.Node
 				continue
 			}
 			capVal := prov.Capacity
-			if capVal <= 0 {
-				capVal = concurrencyFallback
-			}
 			inflight, queued := m.getStatsForProviderLocked(nodeID, prov.ID)
 
 			servedModels := make([]string, len(prov.Models))
@@ -581,6 +573,10 @@ func (m *modelQueueManager) getSnapshotForNode(nodeID string, rec *edgenode.Node
 	}
 
 	// Legacy adapter snapshots for nodes with no providers catalog.
+	concurrencyFallback := 1
+	if rec.Runtime.Concurrency > 0 {
+		concurrencyFallback = rec.Runtime.Concurrency
+	}
 
 	// 1. CLI
 	if rec.Adapters.CLI.Enabled {

apps/edge/internal/service/run_dispatch.go

@@ -471,8 +471,9 @@ func providerCanServe(prov config.NodeProviderConf, servedModel string) bool {
 	return false
 }
 
-// providerDispatchable checks whether a provider has a non-empty adapter and
-// valid capacity so it can be used for dispatch.
+// providerDispatchable checks whether a provider has a non-empty adapter.
+// Capacity is checked separately so zero-capacity providers remain visible in
+// status snapshots while unavailable for provider-pool dispatch.
 func providerDispatchable(prov config.NodeProviderConf) bool {
 	return strings.TrimSpace(prov.Adapter) != ""
 }

apps/edge/internal/service/status_provider.go

@@ -72,16 +72,12 @@ func NodeEntrySnapshot(entry *edgenode.NodeEntry) NodeSnapshot {
 // zero inflight/queued for nodes whose config has a providers[] catalog but
 // whose service has no queue (e.g. queue disabled at startup).
 func staticProviderCatalogSnapshots(rec *edgenode.NodeRecord) []*iop.ProviderSnapshot {
-	concurrencyFallback := positiveOr(rec.Runtime.Concurrency, defaultNodeCapacity)
 	snaps := make([]*iop.ProviderSnapshot, 0, len(rec.Providers))
 	for _, prov := range rec.Providers {
 		if prov.ID == "" {
 			continue
 		}
 		capVal := prov.Capacity
-		if capVal <= 0 {
-			capVal = concurrencyFallback
-		}
 		servedModels := make([]string, len(prov.Models))
 		copy(servedModels, prov.Models)
 		lifecycleCaps := make([]string, len(prov.LifecycleCapabilities))

apps/edge/internal/service/status_provider_test.go

@@ -276,6 +276,66 @@ func TestListNodeSnapshotsProviderCatalogSupersedesAdapterSnapshots(t *testing.T
 	}
 }
 
+func TestListNodeSnapshotsCatalogProviderZeroCapacityDoesNotUseRuntimeFallback(t *testing.T) {
+	reg := edgenode.NewRegistry()
+	reg.Register(&edgenode.NodeEntry{NodeID: "node-zero-cap", Alias: "node-zero-cap"})
+
+	store := edgenode.NewNodeStore()
+	rec := &edgenode.NodeRecord{
+		ID: "node-zero-cap",
+		Adapters: config.AdaptersConf{
+			OpenAICompatInstances: []config.OpenAICompatInstanceConf{
+				{Name: "vllm-gpu", Enabled: true, Capacity: 7},
+			},
+		},
+		Providers: []config.NodeProviderConf{
+			{ID: "prov-zero", Adapter: "vllm-gpu", Models: []string{"model-zero"}, Health: "available", Capacity: 0},
+		},
+		Runtime: config.RuntimeConf{Concurrency: 7},
+	}
+	store.Add(rec)
+
+	bus := edgeevents.NewBus()
+	svc := New(reg, bus)
+	svc.SetNodeStore(store)
+
+	snaps := svc.ListNodeSnapshots()
+	if len(snaps) != 1 {
+		t.Fatalf("expected 1 node snapshot, got %d", len(snaps))
+	}
+	if len(snaps[0].ProviderSnapshots) != 1 {
+		t.Fatalf("expected 1 provider snapshot, got %d", len(snaps[0].ProviderSnapshots))
+	}
+	p := snaps[0].ProviderSnapshots[0]
+	if p.Id != "prov-zero" {
+		t.Fatalf("expected provider id prov-zero, got %q", p.Id)
+	}
+	if p.Capacity != 0 {
+		t.Fatalf("provider catalog capacity must not fall back to runtime concurrency: got %d", p.Capacity)
+	}
+	if p.LoadRatio != 0 {
+		t.Fatalf("expected zero load ratio for zero capacity provider, got %.2f", p.LoadRatio)
+	}
+}
+
+func TestStaticProviderCatalogSnapshotsZeroCapacityDoesNotUseRuntimeFallback(t *testing.T) {
+	rec := &edgenode.NodeRecord{
+		ID: "node-static-zero-cap",
+		Providers: []config.NodeProviderConf{
+			{ID: "prov-zero", Adapter: "vllm-gpu", Models: []string{"model-zero"}, Health: "available", Capacity: 0},
+		},
+		Runtime: config.RuntimeConf{Concurrency: 7},
+	}
+
+	snaps := staticProviderCatalogSnapshots(rec)
+	if len(snaps) != 1 {
+		t.Fatalf("expected 1 provider snapshot, got %d", len(snaps))
+	}
+	if snaps[0].Capacity != 0 {
+		t.Fatalf("static provider catalog capacity must not fall back to runtime concurrency: got %d", snaps[0].Capacity)
+	}
+}
+
 // TestListNodeSnapshotsLegacyAdapterFallbackWhenNoProviders verifies that when
 // a node has no providers[] catalog, ListNodeSnapshots returns the legacy adapter
 // snapshots (CLI, Ollama, etc.) rather than empty results.

configs/edge.yaml

@@ -128,7 +128,7 @@ openai:
   # nodes[].providers[].adapter — concrete adapter instance key used by edge to dispatch.
   # nodes[].providers[].models — served model names this provider can serve.
   # nodes[].providers[].health — observed health state string.
-  # nodes[].providers[].capacity — max concurrent execution slots.
+  # nodes[].providers[].capacity — provider-pool max concurrent execution slots; 0 is not dispatchable.
   # nodes[].providers[].max_queue — max queue depth (per-provider).
   # nodes[].providers[].queue_timeout_ms — queue timeout in milliseconds.
   # nodes[].providers[].lifecycle_capabilities — coarse lifecycle capabilities list.

packages/go/config/config.go

@@ -128,7 +128,9 @@ type NodeProviderConf struct {
 	Models []string `mapstructure:"models" yaml:"models,omitempty"`
 	// Health is the observed provider health state.
 	Health string `mapstructure:"health" yaml:"health,omitempty"`
-	// Capacity is the maximum concurrent execution slots.
+	// Capacity is the provider/resource maximum concurrent execution slots.
+	// Provider-pool dispatch treats 0 as not dispatchable rather than falling
+	// back to node runtime concurrency metadata.
 	Capacity int `mapstructure:"capacity" yaml:"capacity,omitempty"`
 	// MaxQueue is the maximum queue depth.
 	MaxQueue int `mapstructure:"max_queue" yaml:"max_queue,omitempty"`