fix(apiserver): apikeys whitelist + job_status type + mcp_operator queue:control

Three fixes found during scheduler-agent Phase 1 cluster verification:

- whitelist.go: add /xce/v1/apikeys to WhitelistPOSTURL (Phase 0 added the
  route but never whitelisted it → 403 "URL is not allowed"). bug-1489.
- jobs.go: fix job_status int64 assertion in job control. store.Jobs returns
  status as int/float64; the `.(int64)` assertion broke control with
  "job_status is not an int64". Add normalizeJobStatus() accepting int/int32/
  int64/float64/json.Number.
- rbac.yaml: mcp_operator role gains queue:control (the job-control handler
  also checks queue:control on the job's queue); add rbac_mcp_operator_test.

Co-Authored-By: Claude <noreply@anthropic.com>

src/frontend/apiserver/config/rbac.yaml

@@ -171,4 +171,36 @@ roles:
         domain:
           - type: all
 
-
+  # mcp_operator: service-account role for the scheduler-agent MCP server
+  # (Phase 1, write operations). Same cluster visibility as mcp_readonly,
+  # plus queue:submit (scheduler_submit) and job:control (scheduler_cancel/
+  # hold/release). queue:control is required by the apiserver's job-control
+  # handler (jobs.go checks queue:control on the job's queue too), so it is
+  # granted here. An MCP server credential using this role can mutate job
+  # state, but still cannot manage other users/roles or cluster config.
+  mcp_operator:
+    permissions:
+      - action: job:view
+        domain:
+          - type: all
+      - action: queue:view
+        domain:
+          - type: all
+      - action: host:view
+        domain:
+          - type: all
+      - action: resource:view
+        domain:
+          - type: all
+      - action: cluster:view
+        domain:
+          - type: all
+      - action: queue:submit
+        domain:
+          - type: all
+      - action: queue:control
+        domain:
+          - type: all
+      - action: job:control
+        domain:
+          - type: all

src/frontend/apiserver/controllers/jobs.go

@@ -3011,10 +3011,14 @@ func validateJobControl(jobinfo utils.OrderMap, action string,
 	}
 
 	if rawStatus, exists := jobinfo.Data["job_status"]; exists {
-		if jobStatus, ok := rawStatus.(int64); ok {
-			db_job_status := int(jobStatus)
+		// job_status arrives as int (from store.Jobs), int64, or float64
+		// (from JSON decode paths). Normalize before use; the old `.(int64)`
+		// assertion broke control for every job whose status decoded as int
+		// or float64 ("job_status is not an int64").
+		if jobStatusInt, ok := normalizeJobStatus(rawStatus); ok {
+			db_job_status := jobStatusInt
 			job_status := models.JobStateStr(db_job_status)
-			jobStatus = int64(models.JOBSTAT[job_status])
+			jobStatus := int64(models.JOBSTAT[job_status])
 			if utils.IsControlActionAllowed(actionType, jobStatus, models.JobActionControlMap) {
 				if action == "kill" && signal != "" {
 					if !utils.IsControlActionAllowed(actionType, jobStatus, models.JobKillSignalControlMap) {
@@ -3095,6 +3099,26 @@ func foundJob(jobid int64, array_idx int, jobinfos []utils.OrderMap) ([]utils.Or
 	}
 }
 
+// normalizeJobStatus converts a job_status value (int, int64, or float64 from
+// JSON decode) into a plain int. Returns ok=false for unsupported types.
+func normalizeJobStatus(v any) (int, bool) {
+	switch s := v.(type) {
+	case int:
+		return s, true
+	case int32:
+		return int(s), true
+	case int64:
+		return int(s), true
+	case float64:
+		return int(s), true
+	case json.Number:
+		if n, err := s.Int64(); err == nil {
+			return int(n), true
+		}
+	}
+	return 0, false
+}
+
 func processJobData(rows []utils.OrderMap, pagetoken string) utils.QueryResult {
 	logger.Debug("Enter processJobData() function", zap.Any("rows", rows))
 	for _, row := range rows {

src/frontend/apiserver/controllers/rbac_mcp_operator_test.go

@@ -0,0 +1,55 @@
+package controllers
+
+import (
+	"apiserver/store"
+	"os"
+	"path/filepath"
+	"testing"
+)
+
+// TestRBACYAML_McpOperatorRole_LoadsWritePermissions verifies the
+// mcp_operator role (added for the scheduler-agent MCP server's service
+// account, Phase 1) is present in the shipped config/rbac.yaml and grants the
+// write permissions the Phase 1 tools need: queue:submit (scheduler_submit)
+// and job:control (scheduler_cancel/hold/release) — all on domain type "all".
+func TestRBACYAML_McpOperatorRole_LoadsWritePermissions(t *testing.T) {
+	resetRBACState()
+
+	path, err := filepath.Abs("../config/rbac.yaml")
+	if err != nil {
+		t.Fatalf("resolve rbac.yaml path: %v", err)
+	}
+	if _, err := os.Stat(path); err != nil {
+		t.Skipf("config/rbac.yaml not found at %s: %v", path, err)
+	}
+
+	if err := LoadRBACConfig(path); err != nil {
+		t.Fatalf("LoadRBACConfig: %v", err)
+	}
+
+	if !store.Roles.Has("mcp_operator") {
+		t.Fatal("expected mcp_operator role to be loaded from config/rbac.yaml")
+	}
+
+	perms := store.GetRolePermissions("mcp_operator")
+	if len(perms) == 0 {
+		t.Fatal("mcp_operator should have at least one permission")
+	}
+
+	// Write permissions required by the Phase 1 tools must be present and
+	// scoped to "all" domains.
+	writePerms := map[string]bool{
+		"queue:submit": false,
+		"job:control":  false,
+	}
+	for _, p := range perms {
+		if _, needed := writePerms[p.Action]; needed && p.DomainType == "all" {
+			writePerms[p.Action] = true
+		}
+	}
+	for action, found := range writePerms {
+		if !found {
+			t.Errorf("mcp_operator should grant %s with domain type all", action)
+		}
+	}
+}

src/frontend/apiserver/middleware/whitelist.go

@@ -36,6 +36,7 @@ var WhitelistGETURL = []string{
 var WhitelistPOSTURL = []string{
 	"/xce/v1/auth/logon",     // logon
 	"/xce/v1/auth/logout",    // logout
+	"/xce/v1/apikeys",        // generate service-account API key (Phase 0)
 	"/xce/v1/hosts/control",  // control hosts
 	"/xce/v1/hosts/select",   // filter host with -R select string
 	"/xce/v1/queues/control", // control queues