diff --git a/wwwroot/api/cluster_nodes.dspy b/wwwroot/api/cluster_nodes.dspy
new file mode 100644
index 0000000..77bacef
--- /dev/null
+++ b/wwwroot/api/cluster_nodes.dspy
@@ -0,0 +1,26 @@
+# 查询集群已分配节点
+# params: cluster_id (必填)
+cluster_id = params_kw.get('cluster_id', '')
+if not cluster_id:
+    return {'status': 'error', 'message': 'Missing cluster_id'}
+
+env = request._run_ns
+dbname = env.get_module_dbname('pccs')
+async with DBPools().sqlorContext(dbname) as sor:
+    rows = await sor.sqlExe(
+        "SELECT cn.id as node_id, cn.name, cn.ip_address, cn.cpu_cores, cn.memory_gb, "
+        "cn.gpu_count, cn.gpu_type, cnn.role, cnn.status as node_status, cnn.assigned_at "
+        "FROM cluster_node cnn "
+        "JOIN compute_node cn ON cn.id = cnn.node_id "
+        "WHERE cnn.cluster_id = ${cid}$ AND cnn.status != 'removed' "
+        "ORDER BY cnn.role, cn.name",
+        {'cid': cluster_id}
+    )
+    result = [{
+        'node_id': r.node_id, 'name': r.name, 'ip_address': r.ip_address,
+        'cpu_cores': r.cpu_cores, 'memory_gb': r.memory_gb,
+        'gpu_count': r.gpu_count, 'gpu_type': r.gpu_type,
+        'role': r.role, 'status': r.node_status,
+        'assigned_at': str(r.assigned_at)
+    } for r in rows]
+    return {'status': 'ok', 'data': result, 'total': len(result)}
diff --git a/wwwroot/api/node_allocate.dspy b/wwwroot/api/node_allocate.dspy
index 74a62d2..005c3b5 100644
--- a/wwwroot/api/node_allocate.dspy
+++ b/wwwroot/api/node_allocate.dspy
@@ -1,4 +1,28 @@
-# 从算力池分配节点到集群
-# params: pool_id, cluster_id, role(control/compute), count, cpu, memory, gpu
+# 从算力池分配节点到集群 (算力单元分配)
+# params:
+#   pool_id    - 算力池 ID (必填)
+#   cluster_id - 目标集群 ID
+#   role       - 节点角色: control/compute (默认 compute)
+#   count      - 分配数量 (默认 1)
+#   cpu        - 每节点 CPU 核数要求 (可选，0 表示不限制)
+#   memory     - 每节点内存 GB 要求 (可选)
+#   gpu        - 每节点 GPU 数量要求 (可选)
+# return: {status, allocated_nodes: [node_id, ...]}
 result = await allocate_nodes(request, params_kw)
-return result if isinstance(result, dict) else {'status': 'error', 'message': str(result)}
+if isinstance(result, dict) and result.get('status') == 'ok':
+    return {
+        'widgettype': 'Message',
+        'options': {
+            'title': '分配成功',
+            'message': '已分配 {} 个节点:\n{}'.format(
+                len(result.get('allocated_nodes', [])),
+                ', '.join(result.get('allocated_nodes', []))
+            ),
+            'type': 'success',
+            'timeout': 5
+        }
+    }
+return result if isinstance(result, dict) else {
+    'widgettype': 'Error',
+    'options': {'title': '分配失败', 'message': str(result), 'cwidth': 16, 'cheight': 9, 'timeout': 3}
+}
diff --git a/wwwroot/api/node_list_available.dspy b/wwwroot/api/node_list_available.dspy
new file mode 100644
index 0000000..db973b4
--- /dev/null
+++ b/wwwroot/api/node_list_available.dspy
@@ -0,0 +1,24 @@
+# 查询可用算力节点
+# params: pool_id (必填), role (可选 control/compute)
+pool_id = params_kw.get('pool_id', '')
+role = params_kw.get('role', '')
+if not pool_id:
+    return {'status': 'error', 'message': 'Missing pool_id'}
+
+env = request._run_ns
+dbname = env.get_module_dbname('pccs')
+async with DBPools().sqlorContext(dbname) as sor:
+    sql = "SELECT * FROM compute_node WHERE pool_id = ${pid}$ AND status = 'available'"
+    params = {'pid': pool_id}
+    if role:
+        sql = sql + " AND node_type = ${role}$"
+        params['role'] = role
+    sql = sql + " ORDER BY cpu_cores DESC, memory_gb DESC"
+    rows = await sor.sqlExe(sql, params)
+    result = [{
+        'id': r.id, 'name': r.name, 'ip_address': r.ip_address,
+        'cpu_cores': r.cpu_cores, 'memory_gb': r.memory_gb,
+        'gpu_count': r.gpu_count, 'gpu_type': r.gpu_type,
+        'node_type': r.node_type
+    } for r in rows]
+    return {'status': 'ok', 'data': result, 'total': len(result)}
diff --git a/wwwroot/api/node_release.dspy b/wwwroot/api/node_release.dspy
index a1a0bc7..b3ded12 100644
--- a/wwwroot/api/node_release.dspy
+++ b/wwwroot/api/node_release.dspy
@@ -1,3 +1,20 @@
-# 回收节点回算力池
+# 回收节点回算力池 (算力单元回收)
+# params:
+#   node_ids - 节点 ID 列表，逗号分隔 (必填)
+#   pool_id  - 算力池 ID (可选，不传则自动从节点关联获取)
+# return: {status, released: count}
 result = await release_nodes(request, params_kw)
-return result if isinstance(result, dict) else {'status': 'error', 'message': str(result)}
+if isinstance(result, dict) and result.get('status') == 'ok':
+    return {
+        'widgettype': 'Message',
+        'options': {
+            'title': '回收成功',
+            'message': '已回收 {} 个节点到算力池'.format(result.get('released', 0)),
+            'type': 'success',
+            'timeout': 5
+        }
+    }
+return result if isinstance(result, dict) else {
+    'widgettype': 'Error',
+    'options': {'title': '回收失败', 'message': str(result), 'cwidth': 16, 'cheight': 9, 'timeout': 3}
+}
diff --git a/wwwroot/api/pool_stats.dspy b/wwwroot/api/pool_stats.dspy
new file mode 100644
index 0000000..fc01364
--- /dev/null
+++ b/wwwroot/api/pool_stats.dspy
@@ -0,0 +1,4 @@
+# 算力池资源统计
+# params: pool_id (可选，不传则返回全部池)
+result = await pool_stats(request, params_kw)
+return result