feat: 实现 Instance Ready 领域判定与恢复规则

Merge pull request 'feat: 迁移 Database Instance 领域基线与扩展观测' (#4 ) from feat/database-instance-domain into main
Reviewed-on: #4
2026-09-21 07:49:41 +00:00 · 2026-09-21 07:42:08 +00:00 · 2026-09-21 07:33:10 +00:00 · 2026-09-21 06:32:40 +00:00 · 2026-09-21 06:32:40 +00:00 · 2026-09-21 06:32:19 +00:00
16 changed files with 757 additions and 1132 deletions
@@ -0,0 +1,76 @@
+---
+name: homelab-knowledge
+description: Query and maintain the shared homelab-wiki when working on homelab services, infrastructure, architecture, operations, or current service status. Use it to gather existing context before work and to keep durable knowledge synchronized after relevant changes; do not use it for unrelated software work or as a substitute for commit and PR history.
+---
+
+# Homelab Knowledge
+
+Use `homelab-wiki` as the shared long-lived knowledge base for people and agents. Search it directly with `rg`; do not introduce a search index, vector database, or generated copy of the wiki.
+
+## Locate the wiki
+
+Resolve the checkout in this order:
+
+1. `$HOMELAB_WIKI_PATH`, when set.
+2. A sibling directory named `homelab-wiki` next to the current repository.
+3. `/home/panxiao81/homelab-wiki` when it exists.
+
+If no checkout is available, report that constraint. Do not silently skip the knowledge step, clone a repository, or create a replacement wiki without the user's authorization.
+
+Before using the wiki, read its `AGENTS.md` completely. For edits, also read `README.md` and `CONTRIBUTING.md` completely and follow any more specific instructions associated with the target page.
+
+## Gather context
+
+At the beginning of a homelab task:
+
+1. Derive search terms from the component name, service aliases, hostnames, Kubernetes resources, configuration keys, error text, and task intent.
+2. Use `rg -n -i` in the wiki to find candidate pages. Prefer several precise searches over reading the whole repository.
+3. Follow the wiki's task index, service index, architecture constraints, source records, and verification conflicts when they are relevant.
+4. Read the closest authoritative pages and their material links before making decisions. Also read the corresponding source repository README or runbook when changing an implementation.
+5. Distinguish documented design, declared configuration, deployment history, live verification, and work currently in progress. Do not present one as another.
+
+For questions about current project or service status, first obtain the maintainer's current-work and ticket context as required by the wiki, unless the conversation already provides that authorization and scope. Reading documentation does not authorize live-system inspection.
+
+Answer read-only questions from the evidence found. Include paths or links that let the user verify important claims, and state when evidence may be stale or conflicting.
+
+## Maintain knowledge after changes
+
+For any code, configuration, infrastructure, or operational change, perform a documentation-impact check before declaring the task complete.
+
+Update the wiki in the same task when the change affects durable knowledge such as:
+
+- service purpose, lifecycle, entry point, authentication, permissions, dependencies, or first-use path;
+- architecture boundaries or accepted constraints;
+- deployment ownership or persistent operating behavior;
+- troubleshooting, recovery, verification, or maintenance procedures;
+- the addition, replacement, or retirement of a service.
+
+Keep one-time progress, implementation narration, and release-by-release history in commits, PRs, or tickets. Do not copy them into the wiki unless they change a durable stage summary. Implementation-specific parameters may remain in the source repository README or runbook when the wiki convention says to link rather than duplicate them.
+
+When editing:
+
+1. Inspect both the source-repository diff and the wiki working tree before writing. Preserve unrelated user changes in both repositories.
+2. Update the page closest to the fact first, then only the navigation, indexes, constraints, or verification records that the wiki rules require.
+3. Preserve evidence metadata. Never advance `last_verified` without performing the stated live verification; ordinary review may update only fields permitted by the wiki.
+4. Link related source commits, PRs, or paths when available. Clearly mark uncommitted sources and unfinished cross-repository synchronization.
+5. Record conflicts rather than resolving them by assumption. Ask before live inspection or before choosing among materially conflicting current-state claims.
+6. Keep credentials, tokens, private keys, Terraform state, secret values, and sensitive command output out of documentation. Never read or copy known sensitive files merely to improve the wiki.
+
+Wiki edits are a separate repository change. Do not commit, push, open a PR, or modify a live system unless the user has authorized that action.
+
+## Verify and report
+
+After editing the wiki, run from its root:
+
+```bash
+python3 scripts/check_docs.py
+git diff --check
+```
+
+If the checker itself changed, also run:
+
+```bash
+python3 -m unittest discover -s tests -v
+```
+
+In the final response, report source-repository changes and wiki changes separately, including validation performed and anything still awaiting verification or cross-repository linkage. If no wiki update was needed, state the concrete reason; do not merely say that documentation was unaffected.
@@ -20,7 +20,6 @@ import (
 	"sigs.k8s.io/controller-runtime/pkg/webhook"

 	executionv1alpha1 "git.ddupan.top/panxiao81/ayatori/api/execution/v1alpha1"
-	"git.ddupan.top/panxiao81/ayatori/internal/controller"
 	// +kubebuilder:scaffold:imports
 )

@@ -166,13 +165,6 @@ func main() {
 		os.Exit(1)
 	}

-	if err := (&controller.JobReconciler{
-		Client: mgr.GetClient(),
-	}).SetupWithManager(mgr); err != nil {
-		setupLog.Error(err, "Failed to create controller", "controller", "Job")
-		os.Exit(1)
-	}
-
 	// +kubebuilder:scaffold:builder

 	if err := mgr.AddHealthzCheck("healthz", healthz.Ping); err != nil {
@@ -1,58 +1,11 @@
---
 apiVersion: rbac.authorization.k8s.io/v1
 kind: ClusterRole
 metadata:
+  labels:
+    app.kubernetes.io/name: ayatori
+    app.kubernetes.io/managed-by: kustomize
  name: manager-role
 rules:
- apiGroups:
-  - ""
-  resources:
-  - namespaces
-  - serviceaccounts
-  verbs:
-  - get
-  - list
-  - watch
- apiGroups:
-  - batch
-  resources:
-  - jobs
-  verbs:
-  - create
-  - delete
-  - get
-  - list
-  - watch
- apiGroups:
-  - execution.ayatori.ddupan.top
-  resources:
-  - jobclasses
-  - kubernetesexecutionparameters
-  verbs:
-  - get
-  - list
-  - watch
- apiGroups:
-  - execution.ayatori.ddupan.top
-  resources:
-  - jobs
-  verbs:
-  - get
-  - list
-  - patch
-  - update
-  - watch
- apiGroups:
-  - execution.ayatori.ddupan.top
-  resources:
-  - jobs/finalizers
-  verbs:
-  - update
- apiGroups:
-  - execution.ayatori.ddupan.top
-  resources:
-  - jobs/status
-  verbs:
-  - get
-  - patch
-  - update
+- apiGroups: [""]
+  resources: ["pods"]
+  verbs: ["get", "list", "watch"]
@@ -21,15 +21,17 @@ Database 是 Ayatori 首批实际产品领域之一。第一个迁移切片只

 代码被移动到 Ayatori 的 `internal/database/domain/instance`，测试 import 和文档链接相应更新；
 首个后续切片按已批准合同增加 Instance extension observation：观测与当前 target 绑定，进入重新
-验证或删除时失效，且支持判定不授权 Tenant provisioning。其余 Ready/observation 行为仍应先更新
-合同与测试再实现，不能把旧运行链路接回该模型。
+验证或删除时失效，且支持判定不授权 Tenant provisioning。后续 Ready 切片实现管理能力判定、
+registry 准备决策与完整回读、Ready 重验及本轮 evidence 前置检查；沿用已批准合同，不能把旧运行
+链路接回该模型。各层验证边界见 [Instance 领域规格](domain-instance.md)。

 ## 边界

 - 领域层不依赖 Kubernetes types、数据库 driver 或凭据 provider。
 - CredentialReference 只携带管理 Secret 的名称与字段映射，不包含 Secret 内容或 OpenBao path。
 - Instance checkpoint 不是外部事实；实际能力必须由 application/adapter 观察后交给领域对象判断。
- 当前代码不授权 Tenant provisioning，也不表示 Database API 已经可用。
+- 当前代码只检查 Instance 供应前置条件，不授予 Tenant 所有权或外部写入权限，也不表示
+  Database API 已经可用。

 ## 设计入口

@@ -2,7 +2,7 @@

 状态：Draft，含已确认决策。日期：2026-09-13。

-上层合并边界见 [ADR-0008](../decisions/0008-merge-Ayatori Database controller.md)。本文只展开 Instance，不包含 Tenant 的供应
+上层合并边界见 [ADR-0008](../decisions/0008-merge-postgresql-tenant-operator.md)。本文只展开 Instance，不包含 Tenant 的供应
 实现，也不新增 CRD 字段。设计签名用于评审职责与行为，不是待复制的 Go 接口代码。

 ## 1. 对象职责与生命周期
@@ -228,3 +228,26 @@ Ready --registry 需修复/保存--> InitializingRegistry
 均已确认。其他决策及未决项见总体草案，不增加后台清扫器或状态字段。

 批准本对象结构不等于批准这些未决行为，也不意味着立刻实现完整供应链路。
+
+## 8. 领域实现与验证边界（2026-09-21）
+
+`internal/database/domain/instance` 按上述方法合同实现 Ready 纯判定。输入分别表达连接、
+metadata、role、database、grant、extension 管理能力，以及 registry 的未观察、缺失、需迁移、
+可用、不兼容和不可访问状态。检查零值或未知值按证据不足处理；操作失败只使用封闭的安全
+类别，不接收驱动错误。`Snapshot.Failure` 是 Condition 映射的领域输入，不新增 CRD/status 字段。
+
+管理能力分别指目标连接可用、服务器 metadata 可读，以及执行规格 §7 所要求的角色、数据库、
+授权和扩展管理操作的能力；不是仅凭版本查询或扩展可用列表判定权限。具体 SQL 权限探测矩阵、
+最小权限角色和扩展权限例外仍须在 PostgreSQL adapter 切片定义并用真实后端验证。
+registry 不兼容独立保留为领域失败类别，不将其误报为权限不足；公开 Condition Reason 的映射
+留待 API/application 切片按原合同评审。
+
+领域测试验证完整回读、缺少检查项、状态重建、重复判定、目标不匹配、配置变化、依赖失败、
+registry 丢失／不兼容、操作结果不确定和删除限制。只有本轮完整能力判定通过后，Instance
+前置条件检查才通过；这不授予 Tenant 所有权，也不替代实际写入前的并发校验。
+
+本切片不新增 controller、adapter、Secret 读取或外部生命周期操作。checkpoint 保存失败、
+resourceVersion 冲突、watch 与 finalizer 事件链需由后续 application/envtest 验证；SQL 探测、
+registry 初始化／迁移、超时后的真实状态回读和并发幂等由 PostgreSQL 集成测试验证；Secret
+变化后的连接刷新由 Kubernetes API 加真实 PostgreSQL 的集成测试验证。纯领域测试不能证明
+Database 已可运行或这些集成合同已完成。
@@ -30,8 +30,12 @@ PostgreSQL Tenant Operator 合并为 Ayatori 的 Database 领域模块。保留

 当前没有可用发布版本、没有被该 operator 托管的 PostgreSQL 实例或 Tenant，也没有需要在线
 转换的已部署 CR。因此此次合并不承担旧实现兼容性：旧运行链路可以直接撤销，不保留直接读取
-OpenBao 管理凭据的路径，不兼容旧 status checkpoint、samples 或落后于规范的 CRD。API 字段若
-妨碍清晰领域模型、恢复行为或测试，可以在 `v1alpha1` 阶段修改并重新生成。
+OpenBao 管理凭据的路径，也不兼容落后于规范的旧 CRD、samples 或实现细节。
+
+没有部署兼容负担不等于重新设计已经批准的产品合同。源项目的系统规格、API 语义、Instance 与
+Tenant 领域模型、状态机、ownership registry、OpenBao/ExternalSecret 凭据交付、Retain/Delete、
+恢复与测试设计整体作为 Ayatori Database 模块的规范基线。除 API group、项目归属和装配结构外，
+迁移不得静默改变这些行为；确需改变时必须先单独修订规格并记录决定。

 目标结构遵守 Ayatori 的模块化单体边界：

@@ -53,14 +57,15 @@ Database API 直接重构为 Ayatori 统一结构：API group 使用
 domain 与 adapter 放入 Ayatori 对应 Database 模块。原 `database.ddupan.top/v1alpha1` 不保留
 别名、conversion 或兼容入口。

-迁移前逐项核对批准规格、领域模型与当前 Go types；冲突时以批准规格和代码质量为基线，并在
-Ayatori 中记录有意改变。无需实现在线 CRD conversion 或数据迁移。
+迁移前逐项核对批准规格、领域模型与当前 Go types；冲突时以批准规格为准。代码质量通过重写
+旧运行链路、清晰 application/adapter 边界和测试实现，不通过改变已批准行为获得。无需实现在线
+CRD conversion 或数据迁移。

 ## 迁移方式

 1. 以包含已合并 Instance 领域基础和 CI #14 的最新 `main` commit 作为 source reference；记录
-   commit，并先提取规范、领域模型和纯单元测试中仍然成立的部分。目标是保留知识与验证，不是
-   逐文件复制旧实现。
+   commit，将完整批准规格与设计文档迁入 Ayatori Database 文档，并迁移领域模型和纯单元测试。
+   设计合同直接复用；旧运行代码不逐文件复制。
 2. 保留源仓库暂停中的脏工作树，不移动、提交或复制两个 extension observation 文件。以后可以
   先在源仓库形成独立 commit，或在 Ayatori 根据批准合同重新实现，但不得把未提交内容描述为来源。
 3. 在 Ayatori multi-group 项目中用 Kubebuilder 注册 Database API，按批准规格迁移 types，重新
@@ -82,5 +87,6 @@ Ayatori 中记录有意改变。无需实现在线 CRD conversion 或数据迁
 - 已批准的 DBaaS 设计与测试投资得到保留。
 - 单一 manager/release 不意味着领域耦合；Database 仍保持独立 package、adapter 和测试边界。
 - 可以从已合并的领域基础开始迁移；旧运行链路和未提交 extension observation 不进入首个切片。
- 无部署兼容负担允许优先修正 API 和架构，不为尚未使用的旧代码保留技术债。
+- 无部署兼容负担允许彻底重写旧运行链路，不为尚未使用的实现技术债保留兼容层；已批准设计合同
+  仍然有效。
 - Database 使用 Ayatori 统一 API group 与目录结构，不为未投入使用的旧 group 保留入口。
@@ -51,9 +51,10 @@
 - 用户集群只暴露 worker node，控制面完全由平台托管。
 - 本节记录候选实现边界，不构成路线图承诺。

-## 首个业务里程碑
+## 后续 Compute 验收场景

-完成 Laptop Rebuild Readiness：
+Database 等首批资源优先落地。Compute 开始实施后，以 Laptop Rebuild Readiness 验证节点
+生命周期与恢复能力；该场景不作为首批 Database、LoadBalancer 或 Bucket 的交付前置条件：

 1. 临时节点加入。
 2. laptop 上的 workload 被重建、迁移或形成可执行人工任务。
@@ -1,132 +0,0 @@
-package kubernetes
-
-import (
-	"fmt"
-
-	executionv1alpha1 "git.ddupan.top/panxiao81/ayatori/api/execution/v1alpha1"
-	batchv1 "k8s.io/api/batch/v1"
-	corev1 "k8s.io/api/core/v1"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/apimachinery/pkg/runtime/schema"
-)
-
-const (
-	ControllerName = "execution.ayatori.ddupan.top/kubernetes"
-	ReferenceType  = "Job"
-	JobUIDLabel    = "execution.ayatori.ddupan.top/job-uid"
-)
-
-var ayatoriJobGVK = schema.GroupVersionKind{
-	Group:   executionv1alpha1.GroupVersion.Group,
-	Version: executionv1alpha1.GroupVersion.Version,
-	Kind:    "Job",
-}
-
-// BuildJob translates the stable execution API into the Kubernetes adapter's
-// backend object. It intentionally does not accept or expose a PodSpec.
-func BuildJob(
-	job *executionv1alpha1.Job,
-	parameters *executionv1alpha1.KubernetesExecutionParameters,
-	resources executionv1alpha1.ExecutionResourceRequirements,
-) *batchv1.Job {
-	backoffLimit := int32(0)
-	controller := true
-	blockOwnerDeletion := true
-
-	//nolint:modernize // ObjectMeta is promoted through embedded TypeMeta; embedlit produces invalid Go here.
-	return &batchv1.Job{
-		ObjectMeta: metav1.ObjectMeta{
-			Name:      job.Name,
-			Namespace: job.Namespace,
-			Labels: map[string]string{
-				JobUIDLabel: string(job.UID),
-			},
-			OwnerReferences: []metav1.OwnerReference{{
-				APIVersion:         ayatoriJobGVK.GroupVersion().String(),
-				Kind:               ayatoriJobGVK.Kind,
-				Name:               job.Name,
-				UID:                job.UID,
-				Controller:         &controller,
-				BlockOwnerDeletion: &blockOwnerDeletion,
-			}},
-		},
-		Spec: batchv1.JobSpec{
-			BackoffLimit: &backoffLimit,
-			Template: corev1.PodTemplateSpec{
-				ObjectMeta: metav1.ObjectMeta{Labels: map[string]string{JobUIDLabel: string(job.UID)}},
-				Spec: corev1.PodSpec{
-					RestartPolicy:      corev1.RestartPolicyNever,
-					ServiceAccountName: parameters.Spec.ServiceAccountName,
-					RuntimeClassName:   optionalString(parameters.Spec.RuntimeClassName),
-					NodeSelector:       parameters.Spec.Scheduling.NodeSelector,
-					Tolerations:        parameters.Spec.Scheduling.Tolerations,
-					SecurityContext:    parameters.Spec.PodSecurityContext,
-					ImagePullSecrets:   job.Spec.Task.ImagePullSecrets,
-					Containers: []corev1.Container{{
-						Name:            "task",
-						Image:           job.Spec.Task.Image,
-						ImagePullPolicy: parameters.Spec.ImagePullPolicy,
-						Command:         job.Spec.Task.Command,
-						Args:            job.Spec.Task.Args,
-						WorkingDir:      job.Spec.Task.WorkingDir,
-						Env:             environment(job.Spec.Task.Env),
-						Resources:       resourceRequirements(resources),
-					}},
-				},
-			},
-		},
-	}
-}
-
-func ValidateOwnership(owner *executionv1alpha1.Job, backend *batchv1.Job) error {
-	if backend.Labels[JobUIDLabel] != string(owner.UID) {
-		return fmt.Errorf("backend Job %s/%s is not owned by Ayatori Job UID %s", backend.Namespace, backend.Name, owner.UID)
-	}
-	return nil
-}
-
-func environment(values []executionv1alpha1.EnvVar) []corev1.EnvVar {
-	result := make([]corev1.EnvVar, 0, len(values))
-	for _, value := range values {
-		env := corev1.EnvVar{Name: value.Name}
-		if value.Value != nil {
-			env.Value = *value.Value
-		}
-		if value.ValueFrom != nil {
-			env.ValueFrom = &corev1.EnvVarSource{
-				SecretKeyRef:    value.ValueFrom.SecretKeyRef,
-				ConfigMapKeyRef: value.ValueFrom.ConfigMapKeyRef,
-			}
-		}
-		result = append(result, env)
-	}
-	return result
-}
-
-func resourceRequirements(resources executionv1alpha1.ExecutionResourceRequirements) corev1.ResourceRequirements {
-	return corev1.ResourceRequirements{
-		Requests: resourceList(resources.Requests),
-		Limits:   resourceList(resources.Limits),
-	}
-}
-
-func resourceList(values executionv1alpha1.ResourceValues) corev1.ResourceList {
-	result := corev1.ResourceList{}
-	if values.CPU != nil {
-		result[corev1.ResourceCPU] = values.CPU.DeepCopy()
-	}
-	if values.Memory != nil {
-		result[corev1.ResourceMemory] = values.Memory.DeepCopy()
-	}
-	if len(result) == 0 {
-		return nil
-	}
-	return result
-}
-
-func optionalString(value string) *string {
-	if value == "" {
-		return nil
-	}
-	return &value
-}
@@ -1,71 +0,0 @@
-package kubernetes
-
-import (
-	"testing"
-
-	executionv1alpha1 "git.ddupan.top/panxiao81/ayatori/api/execution/v1alpha1"
-	corev1 "k8s.io/api/core/v1"
-	"k8s.io/apimachinery/pkg/api/resource"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/apimachinery/pkg/types"
-)
-
-func TestBuildJob(t *testing.T) {
-	literal := "world"
-	cpuRequest := resource.MustParse("100m")
-	memoryLimit := resource.MustParse("128Mi")
-	//nolint:modernize // ObjectMeta is promoted through embedded TypeMeta; embedlit produces invalid Go here.
-	job := &executionv1alpha1.Job{
-		ObjectMeta: metav1.ObjectMeta{Name: "hello", Namespace: "ci", UID: types.UID("job-uid")},
-		Spec: executionv1alpha1.JobSpec{Task: executionv1alpha1.TaskSpec{
-			Image: "alpine:3.22", Command: []string{"echo"}, Args: []string{"hello"},
-			Env: []executionv1alpha1.EnvVar{
-				{Name: "TARGET", Value: &literal},
-				{Name: "TOKEN", ValueFrom: &executionv1alpha1.EnvVarSource{
-					//nolint:modernize // LocalObjectReference is an embedded Kubernetes API field.
-					SecretKeyRef: &corev1.SecretKeySelector{LocalObjectReference: corev1.LocalObjectReference{Name: "token"}, Key: "value"},
-				}},
-			},
-		}},
-	}
-	parameters := &executionv1alpha1.KubernetesExecutionParameters{Spec: executionv1alpha1.KubernetesExecutionParametersSpec{
-		ServiceAccountName: "runner", RuntimeClassName: "runc", ImagePullPolicy: corev1.PullIfNotPresent,
-		Scheduling: executionv1alpha1.KubernetesSchedulingParameters{NodeSelector: map[string]string{"role": "execution"}},
-	}}
-	resources := executionv1alpha1.ExecutionResourceRequirements{
-		Requests: executionv1alpha1.ResourceValues{CPU: &cpuRequest},
-		Limits:   executionv1alpha1.ResourceValues{Memory: &memoryLimit},
-	}
-
-	backend := BuildJob(job, parameters, resources)
-	pod := backend.Spec.Template.Spec
-	if backend.Spec.BackoffLimit == nil || *backend.Spec.BackoffLimit != 0 {
-		t.Fatalf("backoffLimit = %v, want 0", backend.Spec.BackoffLimit)
-	}
-	if pod.RestartPolicy != corev1.RestartPolicyNever || pod.ServiceAccountName != "runner" {
-		t.Fatalf("unexpected pod execution policy: %#v", pod)
-	}
-	if pod.RuntimeClassName == nil || *pod.RuntimeClassName != "runc" {
-		t.Fatalf("runtimeClassName = %v, want runc", pod.RuntimeClassName)
-	}
-	container := pod.Containers[0]
-	if container.Resources.Requests.Cpu().Cmp(cpuRequest) != 0 || container.Resources.Limits.Memory().Cmp(memoryLimit) != 0 {
-		t.Fatalf("resources were not mapped: %#v", container.Resources)
-	}
-	if container.Env[1].ValueFrom == nil || container.Env[1].ValueFrom.SecretKeyRef.Name != "token" {
-		t.Fatalf("secret reference was not preserved: %#v", container.Env[1])
-	}
-	if backend.Labels[JobUIDLabel] != "job-uid" || backend.OwnerReferences[0].UID != job.UID {
-		t.Fatalf("ownership identity was not preserved: %#v", backend.ObjectMeta)
-	}
-}
-
-func TestValidateOwnership(t *testing.T) {
-	job := &executionv1alpha1.Job{}
-	job.UID = types.UID("expected")
-	backend := BuildJob(job, &executionv1alpha1.KubernetesExecutionParameters{}, executionv1alpha1.ExecutionResourceRequirements{})
-	backend.Labels[JobUIDLabel] = "different"
-	if err := ValidateOwnership(job, backend); err == nil {
-		t.Fatal("ValidateOwnership() succeeded for a different Job UID")
-	}
-}
@@ -1,397 +0,0 @@
-package controller
-
-import (
-	"context"
-	"fmt"
-	"slices"
-	"time"
-
-	executionv1alpha1 "git.ddupan.top/panxiao81/ayatori/api/execution/v1alpha1"
-	kubernetesadapter "git.ddupan.top/panxiao81/ayatori/internal/adapter/kubernetes"
-	batchv1 "k8s.io/api/batch/v1"
-	corev1 "k8s.io/api/core/v1"
-	apierrors "k8s.io/apimachinery/pkg/api/errors"
-	"k8s.io/apimachinery/pkg/api/meta"
-	"k8s.io/apimachinery/pkg/api/resource"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/apimachinery/pkg/labels"
-	"k8s.io/apimachinery/pkg/types"
-	ctrl "sigs.k8s.io/controller-runtime"
-	"sigs.k8s.io/controller-runtime/pkg/client"
-	"sigs.k8s.io/controller-runtime/pkg/log"
-)
-
-const (
-	jobFinalizer        = "execution.ayatori.ddupan.top/job-cleanup"
-	reasonResultUnknown = "ResultUnknown"
-)
-
-// JobReconciler executes Ayatori Jobs using supported adapters.
-type JobReconciler struct {
-	client.Client
-	Now func() time.Time
-}
-
-// +kubebuilder:rbac:groups=execution.ayatori.ddupan.top,resources=jobs,verbs=get;list;watch;update;patch
-// +kubebuilder:rbac:groups=execution.ayatori.ddupan.top,resources=jobs/status,verbs=get;update;patch
-// +kubebuilder:rbac:groups=execution.ayatori.ddupan.top,resources=jobs/finalizers,verbs=update
-// +kubebuilder:rbac:groups=execution.ayatori.ddupan.top,resources=jobclasses;kubernetesexecutionparameters,verbs=get;list;watch
-// +kubebuilder:rbac:groups=batch,resources=jobs,verbs=get;list;watch;create;delete
-// +kubebuilder:rbac:groups="",resources=namespaces;serviceaccounts,verbs=get;list;watch
-
-func (r *JobReconciler) Reconcile(ctx context.Context, request ctrl.Request) (ctrl.Result, error) {
-	logger := log.FromContext(ctx)
-	job := &executionv1alpha1.Job{}
-	if err := r.Get(ctx, request.NamespacedName, job); err != nil {
-		return ctrl.Result{}, client.IgnoreNotFound(err)
-	}
-
-	if !job.DeletionTimestamp.IsZero() {
-		return ctrl.Result{}, r.finalize(ctx, job)
-	}
-	if isTerminal(job) {
-		return ctrl.Result{}, nil
-	}
-	if job.Spec.DesiredState == executionv1alpha1.JobDesiredStateCancelled {
-		return ctrl.Result{}, r.cancel(ctx, job)
-	}
-
-	if !containsString(job.Finalizers, jobFinalizer) {
-		job.Finalizers = append(job.Finalizers, jobFinalizer)
-		if err := r.Update(ctx, job); err != nil {
-			return ctrl.Result{}, err
-		}
-		return ctrl.Result{}, nil
-	}
-	if job.Status.Execution != nil {
-		return ctrl.Result{}, r.observeExisting(ctx, job)
-	}
-
-	class, parameters, resources, waiting, err := r.resolve(ctx, job)
-	if err != nil {
-		return ctrl.Result{}, err
-	}
-	if waiting {
-		return ctrl.Result{RequeueAfter: 30 * time.Second}, nil
-	}
-
-	backend := &batchv1.Job{}
-	key := types.NamespacedName{Namespace: job.Namespace, Name: job.Name}
-	err = r.Get(ctx, key, backend)
-	if apierrors.IsNotFound(err) {
-		backend = kubernetesadapter.BuildJob(job, parameters, resources)
-		if err := r.Create(ctx, backend); err != nil {
-			return ctrl.Result{}, err
-		}
-		logger.Info("Created Kubernetes backend Job", "backend", key)
-		return ctrl.Result{}, r.markScheduled(ctx, job, class, parameters, resources, backend)
-	}
-	if err != nil {
-		return ctrl.Result{}, err
-	}
-	if err := kubernetesadapter.ValidateOwnership(job, backend); err != nil {
-		return ctrl.Result{}, r.setCondition(ctx, job, metav1.Condition{
-			Type: executionv1alpha1.JobConditionScheduled, Status: metav1.ConditionFalse,
-			Reason: "BackendConflict", Message: err.Error(),
-		})
-	}
-	if !conditionTrue(job.Status.Conditions, executionv1alpha1.JobConditionScheduled) {
-		return ctrl.Result{}, r.markScheduled(ctx, job, class, parameters, resources, backend)
-	}
-	return ctrl.Result{}, r.observe(ctx, job, backend)
-}
-
-func (r *JobReconciler) observeExisting(ctx context.Context, job *executionv1alpha1.Job) error {
-	if job.Status.Execution.Adapter != "kubernetes" {
-		return r.setCondition(ctx, job, metav1.Condition{
-			Type: executionv1alpha1.JobConditionSucceeded, Status: metav1.ConditionUnknown,
-			Reason: reasonResultUnknown, Message: fmt.Sprintf("adapter %q is not available", job.Status.Execution.Adapter),
-		})
-	}
-	backend := &batchv1.Job{}
-	key := types.NamespacedName{Namespace: job.Namespace, Name: job.Name}
-	if err := r.Get(ctx, key, backend); err != nil {
-		if apierrors.IsNotFound(err) {
-			return r.setCondition(ctx, job, metav1.Condition{
-				Type: executionv1alpha1.JobConditionSucceeded, Status: metav1.ConditionUnknown,
-				Reason: reasonResultUnknown, Message: "Kubernetes backend Job is missing",
-			})
-		}
-		return err
-	}
-	if err := kubernetesadapter.ValidateOwnership(job, backend); err != nil {
-		return r.setCondition(ctx, job, metav1.Condition{
-			Type: executionv1alpha1.JobConditionSucceeded, Status: metav1.ConditionUnknown,
-			Reason: reasonResultUnknown, Message: err.Error(),
-		})
-	}
-	return r.observe(ctx, job, backend)
-}
-
-func (r *JobReconciler) resolve(
-	ctx context.Context,
-	job *executionv1alpha1.Job,
-) (*executionv1alpha1.JobClass, *executionv1alpha1.KubernetesExecutionParameters, executionv1alpha1.ExecutionResourceRequirements, bool, error) {
-	if job.Spec.JobClassName == "" {
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, true, r.reject(ctx, job, "NoDefaultJobClass", "spec.jobClassName is required in the first implementation slice")
-	}
-
-	class := &executionv1alpha1.JobClass{}
-	if err := r.Get(ctx, types.NamespacedName{Name: job.Spec.JobClassName}, class); err != nil {
-		if apierrors.IsNotFound(err) {
-			return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, true, r.reject(ctx, job, "JobClassNotFound", fmt.Sprintf("JobClass %q does not exist", job.Spec.JobClassName))
-		}
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, false, err
-	}
-	if class.Spec.ControllerName != kubernetesadapter.ControllerName {
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, true, r.reject(ctx, job, "UnsupportedController", fmt.Sprintf("controller %q is not supported", class.Spec.ControllerName))
-	}
-	ref := class.Spec.ParametersRef
-	if ref.Group != executionv1alpha1.GroupVersion.Group || ref.Kind != "KubernetesExecutionParameters" {
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, true, r.reject(ctx, job, "InvalidParametersReference", "JobClass must reference KubernetesExecutionParameters")
-	}
-	if allowed, err := r.namespaceAllowed(ctx, job.Namespace, class.Spec.AllowedNamespaces); err != nil {
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, false, err
-	} else if !allowed {
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, true, r.reject(ctx, job, "NamespaceNotAllowed", fmt.Sprintf("namespace %q is not allowed by JobClass %q", job.Namespace, class.Name))
-	}
-
-	parameters := &executionv1alpha1.KubernetesExecutionParameters{}
-	if err := r.Get(ctx, types.NamespacedName{Name: ref.Name}, parameters); err != nil {
-		if apierrors.IsNotFound(err) {
-			return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, true, r.reject(ctx, job, "ParametersNotFound", fmt.Sprintf("KubernetesExecutionParameters %q does not exist", ref.Name))
-		}
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, false, err
-	}
-	serviceAccount := &corev1.ServiceAccount{}
-	if err := r.Get(ctx, types.NamespacedName{Namespace: job.Namespace, Name: parameters.Spec.ServiceAccountName}, serviceAccount); err != nil {
-		if apierrors.IsNotFound(err) {
-			return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, true, r.reject(ctx, job, "ServiceAccountNotFound", fmt.Sprintf("ServiceAccount %q does not exist", parameters.Spec.ServiceAccountName))
-		}
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, false, err
-	}
-
-	resources := applyResourceDefaults(job.Spec.Resources, class.Spec.Resources.Defaults)
-	if err := validateResources(resources); err != nil {
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, true, r.reject(ctx, job, "InvalidResources", err.Error())
-	}
-	if err := r.accept(ctx, job, class, parameters, resources); err != nil {
-		return nil, nil, executionv1alpha1.ExecutionResourceRequirements{}, false, err
-	}
-	return class, parameters, resources, false, nil
-}
-
-func (r *JobReconciler) namespaceAllowed(ctx context.Context, namespace string, selector *metav1.LabelSelector) (bool, error) {
-	if selector == nil {
-		return true, nil
-	}
-	ns := &corev1.Namespace{}
-	if err := r.Get(ctx, types.NamespacedName{Name: namespace}, ns); err != nil {
-		return false, err
-	}
-	compiled, err := metav1.LabelSelectorAsSelector(selector)
-	if err != nil {
-		return false, err
-	}
-	return compiled.Matches(labels.Set(ns.Labels)), nil
-}
-
-func (r *JobReconciler) accept(ctx context.Context, job *executionv1alpha1.Job, class *executionv1alpha1.JobClass, parameters *executionv1alpha1.KubernetesExecutionParameters, resources executionv1alpha1.ExecutionResourceRequirements) error {
-	job.Status.ResolvedJobClass = &executionv1alpha1.ResolvedJobClassReference{
-		Name: class.Name, UID: class.UID, ControllerName: class.Spec.ControllerName,
-		ParametersRef: executionv1alpha1.ParametersReference{
-			Group: class.Spec.ParametersRef.Group, Kind: class.Spec.ParametersRef.Kind,
-			Name: parameters.Name, UID: parameters.UID,
-		},
-	}
-	job.Status.EffectiveResources = resources
-	return r.setCondition(ctx, job, metav1.Condition{
-		Type: executionv1alpha1.JobConditionAccepted, Status: metav1.ConditionTrue,
-		Reason: "Accepted", Message: fmt.Sprintf("JobClass %q accepted", class.Name),
-	})
-}
-
-func (r *JobReconciler) reject(ctx context.Context, job *executionv1alpha1.Job, reason, message string) error {
-	return r.setCondition(ctx, job, metav1.Condition{
-		Type: executionv1alpha1.JobConditionAccepted, Status: metav1.ConditionFalse,
-		Reason: reason, Message: message,
-	})
-}
-
-func (r *JobReconciler) markScheduled(ctx context.Context, job *executionv1alpha1.Job, class *executionv1alpha1.JobClass, parameters *executionv1alpha1.KubernetesExecutionParameters, resources executionv1alpha1.ExecutionResourceRequirements, backend *batchv1.Job) error {
-	job.Status.ResolvedJobClass = &executionv1alpha1.ResolvedJobClassReference{
-		Name: class.Name, UID: class.UID, ControllerName: class.Spec.ControllerName,
-		ParametersRef: executionv1alpha1.ParametersReference{Group: class.Spec.ParametersRef.Group, Kind: class.Spec.ParametersRef.Kind, Name: parameters.Name, UID: parameters.UID},
-	}
-	job.Status.EffectiveResources = resources
-	job.Status.Execution = &executionv1alpha1.ExecutionStatus{
-		Adapter:    "kubernetes",
-		References: []executionv1alpha1.ExecutionReference{{Type: kubernetesadapter.ReferenceType, ID: string(backend.UID)}},
-	}
-	meta.SetStatusCondition(&job.Status.Conditions, condition(job, executionv1alpha1.JobConditionAccepted, metav1.ConditionTrue, "Accepted", "Job accepted"))
-	meta.SetStatusCondition(&job.Status.Conditions, condition(job, executionv1alpha1.JobConditionScheduled, metav1.ConditionTrue, "BackendCreated", "Kubernetes Job created"))
-	meta.SetStatusCondition(&job.Status.Conditions, condition(job, executionv1alpha1.JobConditionSucceeded, metav1.ConditionUnknown, "Pending", "Waiting for task to start"))
-	job.Status.ObservedGeneration = job.Generation
-	return r.Status().Update(ctx, job)
-}
-
-func (r *JobReconciler) observe(ctx context.Context, job *executionv1alpha1.Job, backend *batchv1.Job) error {
-	if job.Status.StartTime == nil && backend.Status.StartTime != nil {
-		job.Status.StartTime = backend.Status.StartTime.DeepCopy()
-	}
-	for _, backendCondition := range backend.Status.Conditions {
-		switch {
-		case backendCondition.Type == batchv1.JobComplete && backendCondition.Status == corev1.ConditionTrue:
-			completion := backend.Status.CompletionTime
-			if completion == nil {
-				now := metav1.NewTime(r.now())
-				completion = &now
-			}
-			job.Status.CompletionTime = completion.DeepCopy()
-			job.Status.Result = &executionv1alpha1.JobResult{Reason: "Completed"}
-			return r.setCondition(ctx, job, metav1.Condition{Type: executionv1alpha1.JobConditionSucceeded, Status: metav1.ConditionTrue, Reason: "Completed", Message: backendCondition.Message})
-		case backendCondition.Type == batchv1.JobFailed && backendCondition.Status == corev1.ConditionTrue:
-			completion := metav1.NewTime(r.now())
-			job.Status.CompletionTime = &completion
-			job.Status.Result = &executionv1alpha1.JobResult{Reason: "ProcessFailed"}
-			return r.setCondition(ctx, job, metav1.Condition{Type: executionv1alpha1.JobConditionSucceeded, Status: metav1.ConditionFalse, Reason: "ProcessFailed", Message: backendCondition.Message})
-		}
-	}
-	reason := "Pending"
-	message := "Waiting for task to start"
-	if backend.Status.StartTime != nil || backend.Status.Active > 0 {
-		reason = "Running"
-		message = "Task is running"
-	}
-	return r.setCondition(ctx, job, metav1.Condition{Type: executionv1alpha1.JobConditionSucceeded, Status: metav1.ConditionUnknown, Reason: reason, Message: message})
-}
-
-func (r *JobReconciler) cancel(ctx context.Context, job *executionv1alpha1.Job) error {
-	backend := &batchv1.Job{}
-	key := types.NamespacedName{Namespace: job.Namespace, Name: job.Name}
-	err := r.Get(ctx, key, backend)
-	if err == nil {
-		if err := kubernetesadapter.ValidateOwnership(job, backend); err != nil {
-			return err
-		}
-		for _, backendCondition := range backend.Status.Conditions {
-			if (backendCondition.Type == batchv1.JobComplete || backendCondition.Type == batchv1.JobFailed) &&
-				backendCondition.Status == corev1.ConditionTrue {
-				return r.observe(ctx, job, backend)
-			}
-		}
-		if err := r.Delete(ctx, backend, client.PropagationPolicy(metav1.DeletePropagationBackground)); err != nil && !apierrors.IsNotFound(err) {
-			return err
-		}
-		return nil
-	}
-	if !apierrors.IsNotFound(err) {
-		return err
-	}
-	now := metav1.NewTime(r.now())
-	job.Status.CompletionTime = &now
-	job.Status.Result = &executionv1alpha1.JobResult{Reason: "Cancelled"}
-	return r.setCondition(ctx, job, metav1.Condition{Type: executionv1alpha1.JobConditionSucceeded, Status: metav1.ConditionFalse, Reason: "Cancelled", Message: "Execution cancelled"})
-}
-
-func (r *JobReconciler) finalize(ctx context.Context, job *executionv1alpha1.Job) error {
-	if !containsString(job.Finalizers, jobFinalizer) {
-		return nil
-	}
-	backend := &batchv1.Job{}
-	key := types.NamespacedName{Namespace: job.Namespace, Name: job.Name}
-	if err := r.Get(ctx, key, backend); err == nil {
-		if err := kubernetesadapter.ValidateOwnership(job, backend); err != nil {
-			return err
-		}
-		if err := r.Delete(ctx, backend, client.PropagationPolicy(metav1.DeletePropagationBackground)); err != nil && !apierrors.IsNotFound(err) {
-			return err
-		}
-		return nil
-	} else if !apierrors.IsNotFound(err) {
-		return err
-	}
-	job.Finalizers = removeString(job.Finalizers, jobFinalizer)
-	return r.Update(ctx, job)
-}
-
-func (r *JobReconciler) setCondition(ctx context.Context, job *executionv1alpha1.Job, next metav1.Condition) error {
-	meta.SetStatusCondition(&job.Status.Conditions, condition(job, next.Type, next.Status, next.Reason, next.Message))
-	job.Status.ObservedGeneration = job.Generation
-	return r.Status().Update(ctx, job)
-}
-
-func condition(job *executionv1alpha1.Job, conditionType string, status metav1.ConditionStatus, reason, message string) metav1.Condition {
-	return metav1.Condition{Type: conditionType, Status: status, Reason: reason, Message: message, ObservedGeneration: job.Generation}
-}
-
-func conditionTrue(conditions []metav1.Condition, conditionType string) bool {
-	current := meta.FindStatusCondition(conditions, conditionType)
-	return current != nil && current.Status == metav1.ConditionTrue
-}
-
-func isTerminal(job *executionv1alpha1.Job) bool {
-	current := meta.FindStatusCondition(job.Status.Conditions, executionv1alpha1.JobConditionSucceeded)
-	return current != nil && (current.Status == metav1.ConditionTrue || current.Status == metav1.ConditionFalse)
-}
-
-func applyResourceDefaults(requested, defaults executionv1alpha1.ExecutionResourceRequirements) executionv1alpha1.ExecutionResourceRequirements {
-	result := requested.DeepCopy()
-	if result.Requests.CPU == nil && defaults.Requests.CPU != nil {
-		result.Requests.CPU = copyQuantity(defaults.Requests.CPU)
-	}
-	if result.Requests.Memory == nil && defaults.Requests.Memory != nil {
-		result.Requests.Memory = copyQuantity(defaults.Requests.Memory)
-	}
-	if result.Limits.CPU == nil && defaults.Limits.CPU != nil {
-		result.Limits.CPU = copyQuantity(defaults.Limits.CPU)
-	}
-	if result.Limits.Memory == nil && defaults.Limits.Memory != nil {
-		result.Limits.Memory = copyQuantity(defaults.Limits.Memory)
-	}
-	return *result
-}
-
-func copyQuantity(value *resource.Quantity) *resource.Quantity {
-	copy := value.DeepCopy()
-	return &copy
-}
-
-func validateResources(resources executionv1alpha1.ExecutionResourceRequirements) error {
-	if resources.Requests.CPU != nil && resources.Limits.CPU != nil && resources.Requests.CPU.Cmp(*resources.Limits.CPU) > 0 {
-		return fmt.Errorf("CPU request must not exceed limit")
-	}
-	if resources.Requests.Memory != nil && resources.Limits.Memory != nil && resources.Requests.Memory.Cmp(*resources.Limits.Memory) > 0 {
-		return fmt.Errorf("memory request must not exceed limit")
-	}
-	return nil
-}
-
-func containsString(values []string, target string) bool {
-	return slices.Contains(values, target)
-}
-
-func removeString(values []string, target string) []string {
-	result := values[:0]
-	for _, value := range values {
-		if value != target {
-			result = append(result, value)
-		}
-	}
-	return result
-}
-
-func (r *JobReconciler) now() time.Time {
-	if r.Now != nil {
-		return r.Now()
-	}
-	return time.Now()
-}
-
-func (r *JobReconciler) SetupWithManager(manager ctrl.Manager) error {
-	return ctrl.NewControllerManagedBy(manager).
-		For(&executionv1alpha1.Job{}).
-		Owns(&batchv1.Job{}).
-		Named("execution-job").
-		Complete(r)
-}
@@ -1,202 +0,0 @@
-package controller
-
-import (
-	"context"
-	"fmt"
-	"os"
-	"path/filepath"
-	"testing"
-	"time"
-
-	executionv1alpha1 "git.ddupan.top/panxiao81/ayatori/api/execution/v1alpha1"
-	kubernetesadapter "git.ddupan.top/panxiao81/ayatori/internal/adapter/kubernetes"
-	batchv1 "k8s.io/api/batch/v1"
-	corev1 "k8s.io/api/core/v1"
-	"k8s.io/apimachinery/pkg/api/meta"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/apimachinery/pkg/runtime"
-	"k8s.io/apimachinery/pkg/types"
-	ctrl "sigs.k8s.io/controller-runtime"
-	"sigs.k8s.io/controller-runtime/pkg/client"
-	"sigs.k8s.io/controller-runtime/pkg/envtest"
-	metricsserver "sigs.k8s.io/controller-runtime/pkg/metrics/server"
-)
-
-const (
-	integrationNamespace = "controller-integration"
-	integrationClass     = "integration"
-)
-
-//nolint:modernize // Kubernetes API structs expose ObjectMeta through embedded TypeMeta fields.
-func TestJobControllerIntegration(t *testing.T) {
-	if os.Getenv("KUBEBUILDER_ASSETS") == "" {
-		t.Skip("KUBEBUILDER_ASSETS is unset; run make test to execute controller integration tests")
-	}
-
-	scheme := runtime.NewScheme()
-	for _, addToScheme := range []func(*runtime.Scheme) error{
-		corev1.AddToScheme,
-		batchv1.AddToScheme,
-		executionv1alpha1.AddToScheme,
-	} {
-		if err := addToScheme(scheme); err != nil {
-			t.Fatal(err)
-		}
-	}
-
-	crdPath, err := filepath.Abs("../../config/crd/bases")
-	if err != nil {
-		t.Fatal(err)
-	}
-	environment := &envtest.Environment{CRDDirectoryPaths: []string{crdPath}}
-	config, err := environment.Start()
-	if err != nil {
-		t.Fatalf("start envtest: %v", err)
-	}
-	t.Cleanup(func() {
-		if err := environment.Stop(); err != nil {
-			t.Errorf("stop envtest: %v", err)
-		}
-	})
-
-	manager, err := ctrl.NewManager(config, ctrl.Options{
-		Scheme:  scheme,
-		Metrics: metricsserver.Options{BindAddress: "0"},
-	})
-	if err != nil {
-		t.Fatal(err)
-	}
-	if err := (&JobReconciler{Client: manager.GetClient()}).SetupWithManager(manager); err != nil {
-		t.Fatal(err)
-	}
-
-	managerContext, cancelManager := context.WithCancel(context.Background())
-	t.Cleanup(cancelManager)
-	managerErrors := make(chan error, 1)
-	go func() {
-		managerErrors <- manager.Start(managerContext)
-	}()
-	if !manager.GetCache().WaitForCacheSync(managerContext) {
-		t.Fatal("manager cache did not synchronize")
-	}
-
-	directClient, err := client.New(config, client.Options{Scheme: scheme})
-	if err != nil {
-		t.Fatal(err)
-	}
-	ctx := context.Background()
-	objects := []client.Object{
-		&corev1.Namespace{ObjectMeta: metav1.ObjectMeta{Name: integrationNamespace, Labels: map[string]string{testLabelKey: testLabelEnabled}}},
-		&corev1.ServiceAccount{ObjectMeta: metav1.ObjectMeta{Name: testSAName, Namespace: integrationNamespace}},
-		&executionv1alpha1.KubernetesExecutionParameters{
-			ObjectMeta: metav1.ObjectMeta{Name: integrationClass},
-			Spec: executionv1alpha1.KubernetesExecutionParametersSpec{
-				ServiceAccountName: testSAName,
-				ImagePullPolicy:    corev1.PullIfNotPresent,
-			},
-		},
-		&executionv1alpha1.JobClass{
-			ObjectMeta: metav1.ObjectMeta{Name: integrationClass},
-			Spec: executionv1alpha1.JobClassSpec{
-				ControllerName: kubernetesadapter.ControllerName,
-				ParametersRef: executionv1alpha1.ParametersReference{
-					Group: executionv1alpha1.GroupVersion.Group,
-					Kind:  "KubernetesExecutionParameters",
-					Name:  integrationClass,
-				},
-				AllowedNamespaces: &metav1.LabelSelector{MatchLabels: map[string]string{testLabelKey: testLabelEnabled}},
-			},
-		},
-	}
-	for _, object := range objects {
-		if err := directClient.Create(ctx, object); err != nil {
-			t.Fatalf("create %T: %v", object, err)
-		}
-	}
-
-	job := &executionv1alpha1.Job{
-		ObjectMeta: metav1.ObjectMeta{Name: testJobName, Namespace: integrationNamespace},
-		Spec: executionv1alpha1.JobSpec{
-			JobClassName: integrationClass,
-			DesiredState: executionv1alpha1.JobDesiredStateRunning,
-			Task:         executionv1alpha1.TaskSpec{Image: "alpine:3.22", Command: []string{"true"}},
-		},
-	}
-	if err := directClient.Create(ctx, job); err != nil {
-		t.Fatal(err)
-	}
-
-	backend := &batchv1.Job{}
-	eventually(t, 10*time.Second, func() (bool, error) {
-		err := directClient.Get(ctx, types.NamespacedName{Namespace: job.Namespace, Name: job.Name}, backend)
-		return err == nil, client.IgnoreNotFound(err)
-	})
-	if backend.Labels[kubernetesadapter.JobUIDLabel] != string(job.UID) {
-		t.Fatalf("backend identity label = %q, want %q", backend.Labels[kubernetesadapter.JobUIDLabel], job.UID)
-	}
-
-	eventually(t, 10*time.Second, func() (bool, error) {
-		if err := directClient.Get(ctx, types.NamespacedName{Namespace: job.Namespace, Name: job.Name}, job); err != nil {
-			return false, err
-		}
-		return conditionStatus(job, executionv1alpha1.JobConditionScheduled) == metav1.ConditionTrue, nil
-	})
-
-	completed := metav1.Now()
-	if err := directClient.Get(ctx, types.NamespacedName{Namespace: job.Namespace, Name: job.Name}, backend); err != nil {
-		t.Fatal(err)
-	}
-	backend.Status.StartTime = &completed
-	backend.Status.CompletionTime = &completed
-	backend.Status.Conditions = []batchv1.JobCondition{
-		{Type: batchv1.JobSuccessCriteriaMet, Status: corev1.ConditionTrue, Reason: "CompletionsReached"},
-		{Type: batchv1.JobComplete, Status: corev1.ConditionTrue, Reason: "Completed"},
-	}
-	if err := directClient.Status().Update(ctx, backend); err != nil {
-		t.Fatal(err)
-	}
-
-	eventually(t, 10*time.Second, func() (bool, error) {
-		if err := directClient.Get(ctx, types.NamespacedName{Namespace: job.Namespace, Name: job.Name}, job); err != nil {
-			return false, err
-		}
-		return conditionStatus(job, executionv1alpha1.JobConditionSucceeded) == metav1.ConditionTrue, nil
-	})
-	if job.Status.StartTime == nil || job.Status.CompletionTime == nil || job.Status.Execution == nil {
-		t.Fatalf("controller did not persist execution status: %#v", job.Status)
-	}
-
-	cancelManager()
-	select {
-	case err := <-managerErrors:
-		if err != nil {
-			t.Fatalf("manager stopped with error: %v", err)
-		}
-	case <-time.After(5 * time.Second):
-		t.Fatal("manager did not stop")
-	}
-}
-
-func eventually(t *testing.T, timeout time.Duration, check func() (bool, error)) {
-	t.Helper()
-	deadline := time.Now().Add(timeout)
-	for time.Now().Before(deadline) {
-		ready, err := check()
-		if err != nil {
-			t.Fatal(err)
-		}
-		if ready {
-			return
-		}
-		time.Sleep(100 * time.Millisecond)
-	}
-	t.Fatal(fmt.Errorf("condition was not met within %s", timeout))
-}
-
-func conditionStatus(job *executionv1alpha1.Job, conditionType string) metav1.ConditionStatus {
-	condition := meta.FindStatusCondition(job.Status.Conditions, conditionType)
-	if condition == nil {
-		return metav1.ConditionUnknown
-	}
-	return condition.Status
-}
@@ -1,245 +0,0 @@
-package controller
-
-import (
-	"context"
-	"testing"
-	"time"
-
-	executionv1alpha1 "git.ddupan.top/panxiao81/ayatori/api/execution/v1alpha1"
-	kubernetesadapter "git.ddupan.top/panxiao81/ayatori/internal/adapter/kubernetes"
-	batchv1 "k8s.io/api/batch/v1"
-	corev1 "k8s.io/api/core/v1"
-	"k8s.io/apimachinery/pkg/api/meta"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/apimachinery/pkg/runtime"
-	"k8s.io/apimachinery/pkg/types"
-	ctrl "sigs.k8s.io/controller-runtime"
-	"sigs.k8s.io/controller-runtime/pkg/client"
-	"sigs.k8s.io/controller-runtime/pkg/client/fake"
-)
-
-const (
-	defaultClassName = "default"
-	testJobName      = "hello"
-	testSAName       = "runner"
-	testLabelKey     = "execution"
-	testLabelEnabled = "enabled"
-)
-
-//nolint:modernize // controller-runtime and Kubernetes API structs expose promoted embedded fields.
-func TestJobReconcilerKubernetesLifecycle(t *testing.T) {
-	ctx := context.Background()
-	now := time.Unix(1_700_000_000, 0)
-	reconciler, kubeClient := testReconciler(t, now, validObjects()...)
-	request := ctrl.Request{}
-	request.NamespacedName = types.NamespacedName{Namespace: "ci", Name: testJobName}
-
-	if _, err := reconciler.Reconcile(ctx, request); err != nil {
-		t.Fatalf("add finalizer: %v", err)
-	}
-	if _, err := reconciler.Reconcile(ctx, request); err != nil {
-		t.Fatalf("create backend: %v", err)
-	}
-
-	backend := &batchv1.Job{}
-	if err := kubeClient.Get(ctx, request.NamespacedName, backend); err != nil {
-		t.Fatalf("backend Job was not created: %v", err)
-	}
-	if backend.Labels[kubernetesadapter.JobUIDLabel] != "ayatori-job-uid" {
-		t.Fatalf("backend UID label = %q", backend.Labels[kubernetesadapter.JobUIDLabel])
-	}
-
-	job := getJob(t, ctx, kubeClient, request.NamespacedName)
-	if !conditionIs(job, executionv1alpha1.JobConditionAccepted, metav1.ConditionTrue) ||
-		!conditionIs(job, executionv1alpha1.JobConditionScheduled, metav1.ConditionTrue) {
-		t.Fatalf("Job was not accepted and scheduled: %#v", job.Status.Conditions)
-	}
-
-	started := metav1.NewTime(now.Add(time.Minute))
-	backend.Status.StartTime = &started
-	backend.Status.Active = 1
-	if err := kubeClient.Status().Update(ctx, backend); err != nil {
-		t.Fatalf("set backend running: %v", err)
-	}
-	if _, err := reconciler.Reconcile(ctx, request); err != nil {
-		t.Fatalf("observe running backend: %v", err)
-	}
-	job = getJob(t, ctx, kubeClient, request.NamespacedName)
-	if job.Status.StartTime == nil || !conditionIs(job, executionv1alpha1.JobConditionSucceeded, metav1.ConditionUnknown) {
-		t.Fatalf("running state was not observed: %#v", job.Status)
-	}
-
-	completed := metav1.NewTime(now.Add(2 * time.Minute))
-	backend = &batchv1.Job{}
-	if err := kubeClient.Get(ctx, request.NamespacedName, backend); err != nil {
-		t.Fatal(err)
-	}
-	backend.Status.Active = 0
-	backend.Status.CompletionTime = &completed
-	backend.Status.Conditions = []batchv1.JobCondition{{Type: batchv1.JobComplete, Status: corev1.ConditionTrue, Reason: "Completed"}}
-	if err := kubeClient.Status().Update(ctx, backend); err != nil {
-		t.Fatalf("set backend complete: %v", err)
-	}
-	if _, err := reconciler.Reconcile(ctx, request); err != nil {
-		t.Fatalf("observe completed backend: %v", err)
-	}
-	job = getJob(t, ctx, kubeClient, request.NamespacedName)
-	if !conditionIs(job, executionv1alpha1.JobConditionSucceeded, metav1.ConditionTrue) || job.Status.CompletionTime == nil {
-		t.Fatalf("terminal state was not observed: %#v", job.Status)
-	}
-}
-
-//nolint:modernize // controller-runtime Request exposes NamespacedName as a promoted embedded field.
-func TestJobReconcilerRejectsMissingClass(t *testing.T) {
-	ctx := context.Background()
-	job := validObjects()[3].(*executionv1alpha1.Job).DeepCopy()
-	job.Spec.JobClassName = "missing"
-	reconciler, kubeClient := testReconciler(t, time.Now(), validObjects()[0], validObjects()[1], job)
-	request := ctrl.Request{}
-	request.NamespacedName = types.NamespacedName{Namespace: job.Namespace, Name: job.Name}
-
-	if _, err := reconciler.Reconcile(ctx, request); err != nil {
-		t.Fatal(err)
-	}
-	result, err := reconciler.Reconcile(ctx, request)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if result.RequeueAfter == 0 {
-		t.Fatal("missing JobClass did not schedule a retry")
-	}
-	stored := getJob(t, ctx, kubeClient, request.NamespacedName)
-	accepted := meta.FindStatusCondition(stored.Status.Conditions, executionv1alpha1.JobConditionAccepted)
-	if accepted == nil || accepted.Status != metav1.ConditionFalse || accepted.Reason != "JobClassNotFound" {
-		t.Fatalf("unexpected Accepted condition: %#v", accepted)
-	}
-}
-
-//nolint:modernize // controller-runtime Request exposes NamespacedName as a promoted embedded field.
-func TestJobReconcilerObservesExistingExecutionWithoutJobClass(t *testing.T) {
-	ctx := context.Background()
-	now := time.Unix(1_700_000_000, 0)
-	job := validObjects()[3].(*executionv1alpha1.Job).DeepCopy()
-	job.Finalizers = []string{jobFinalizer}
-	job.Status.Execution = &executionv1alpha1.ExecutionStatus{Adapter: "kubernetes"}
-	backend := kubernetesadapter.BuildJob(job, validObjects()[4].(*executionv1alpha1.KubernetesExecutionParameters), executionv1alpha1.ExecutionResourceRequirements{})
-	backend.Status.StartTime = &metav1.Time{Time: now}
-	backend.Status.Active = 1
-	reconciler, kubeClient := testReconciler(t, now, job, backend)
-	request := ctrl.Request{NamespacedName: types.NamespacedName{Namespace: job.Namespace, Name: job.Name}}
-
-	if _, err := reconciler.Reconcile(ctx, request); err != nil {
-		t.Fatal(err)
-	}
-	stored := getJob(t, ctx, kubeClient, request.NamespacedName)
-	if stored.Status.StartTime == nil || !conditionIs(stored, executionv1alpha1.JobConditionSucceeded, metav1.ConditionUnknown) {
-		t.Fatalf("existing execution was not observed without its JobClass: %#v", stored.Status)
-	}
-}
-
-//nolint:modernize // controller-runtime Request exposes NamespacedName as a promoted embedded field.
-func TestJobReconcilerCancelsBeforeScheduling(t *testing.T) {
-	ctx := context.Background()
-	job := validObjects()[3].(*executionv1alpha1.Job).DeepCopy()
-	job.Spec.DesiredState = executionv1alpha1.JobDesiredStateCancelled
-	reconciler, kubeClient := testReconciler(t, time.Unix(1_700_000_000, 0), job)
-	request := ctrl.Request{NamespacedName: types.NamespacedName{Namespace: job.Namespace, Name: job.Name}}
-
-	if _, err := reconciler.Reconcile(ctx, request); err != nil {
-		t.Fatal(err)
-	}
-	stored := getJob(t, ctx, kubeClient, request.NamespacedName)
-	condition := meta.FindStatusCondition(stored.Status.Conditions, executionv1alpha1.JobConditionSucceeded)
-	if condition == nil || condition.Status != metav1.ConditionFalse || condition.Reason != "Cancelled" {
-		t.Fatalf("unexpected cancellation condition: %#v", condition)
-	}
-	if stored.Status.CompletionTime == nil {
-		t.Fatal("cancelled Job has no completionTime")
-	}
-}
-
-//nolint:modernize // controller-runtime Request exposes NamespacedName as a promoted embedded field.
-func TestJobReconcilerKeepsConfirmedSuccessDuringCancellation(t *testing.T) {
-	ctx := context.Background()
-	now := time.Unix(1_700_000_000, 0)
-	job := validObjects()[3].(*executionv1alpha1.Job).DeepCopy()
-	job.Spec.DesiredState = executionv1alpha1.JobDesiredStateCancelled
-	job.Finalizers = []string{jobFinalizer}
-	backend := kubernetesadapter.BuildJob(job, validObjects()[4].(*executionv1alpha1.KubernetesExecutionParameters), executionv1alpha1.ExecutionResourceRequirements{})
-	backend.Status.CompletionTime = &metav1.Time{Time: now}
-	backend.Status.Conditions = []batchv1.JobCondition{{Type: batchv1.JobComplete, Status: corev1.ConditionTrue}}
-	reconciler, kubeClient := testReconciler(t, now, job, backend)
-	request := ctrl.Request{NamespacedName: types.NamespacedName{Namespace: job.Namespace, Name: job.Name}}
-
-	if _, err := reconciler.Reconcile(ctx, request); err != nil {
-		t.Fatal(err)
-	}
-	stored := getJob(t, ctx, kubeClient, request.NamespacedName)
-	if !conditionIs(stored, executionv1alpha1.JobConditionSucceeded, metav1.ConditionTrue) {
-		t.Fatalf("confirmed success was overwritten by cancellation: %#v", stored.Status.Conditions)
-	}
-	if err := kubeClient.Get(ctx, request.NamespacedName, &batchv1.Job{}); err != nil {
-		t.Fatalf("successful backend was deleted: %v", err)
-	}
-}
-
-//nolint:modernize // Kubernetes API structs expose ObjectMeta through embedded TypeMeta fields.
-func validObjects() []client.Object {
-	return []client.Object{
-		&corev1.Namespace{ObjectMeta: metav1.ObjectMeta{Name: "ci", Labels: map[string]string{testLabelKey: testLabelEnabled}}},
-		&corev1.ServiceAccount{ObjectMeta: metav1.ObjectMeta{Name: testSAName, Namespace: "ci"}},
-		&executionv1alpha1.JobClass{
-			ObjectMeta: metav1.ObjectMeta{Name: defaultClassName, UID: types.UID("class-uid")},
-			Spec: executionv1alpha1.JobClassSpec{
-				ControllerName:    kubernetesadapter.ControllerName,
-				ParametersRef:     executionv1alpha1.ParametersReference{Group: executionv1alpha1.GroupVersion.Group, Kind: "KubernetesExecutionParameters", Name: defaultClassName},
-				AllowedNamespaces: &metav1.LabelSelector{MatchLabels: map[string]string{testLabelKey: testLabelEnabled}},
-			},
-		},
-		&executionv1alpha1.Job{
-			ObjectMeta: metav1.ObjectMeta{Name: testJobName, Namespace: "ci", UID: types.UID("ayatori-job-uid")},
-			Spec: executionv1alpha1.JobSpec{
-				JobClassName: defaultClassName, DesiredState: executionv1alpha1.JobDesiredStateRunning,
-				Task: executionv1alpha1.TaskSpec{Image: "alpine:3.22", Command: []string{"true"}},
-			},
-		},
-		&executionv1alpha1.KubernetesExecutionParameters{
-			ObjectMeta: metav1.ObjectMeta{Name: defaultClassName, UID: types.UID("parameters-uid")},
-			Spec:       executionv1alpha1.KubernetesExecutionParametersSpec{ServiceAccountName: testSAName, ImagePullPolicy: corev1.PullIfNotPresent},
-		},
-	}
-}
-
-func testReconciler(t *testing.T, now time.Time, objects ...client.Object) (*JobReconciler, client.Client) {
-	t.Helper()
-	scheme := runtime.NewScheme()
-	if err := corev1.AddToScheme(scheme); err != nil {
-		t.Fatal(err)
-	}
-	if err := batchv1.AddToScheme(scheme); err != nil {
-		t.Fatal(err)
-	}
-	if err := executionv1alpha1.AddToScheme(scheme); err != nil {
-		t.Fatal(err)
-	}
-	kubeClient := fake.NewClientBuilder().
-		WithScheme(scheme).
-		WithStatusSubresource(&executionv1alpha1.Job{}, &batchv1.Job{}).
-		WithObjects(objects...).
-		Build()
-	return &JobReconciler{Client: kubeClient, Now: func() time.Time { return now }}, kubeClient
-}
-
-func getJob(t *testing.T, ctx context.Context, kubeClient client.Client, key types.NamespacedName) *executionv1alpha1.Job {
-	t.Helper()
-	job := &executionv1alpha1.Job{}
-	if err := kubeClient.Get(ctx, key, job); err != nil {
-		t.Fatal(err)
-	}
-	return job
-}
-
-func conditionIs(job *executionv1alpha1.Job, conditionType string, status metav1.ConditionStatus) bool {
-	condition := meta.FindStatusCondition(job.Status.Conditions, conditionType)
-	return condition != nil && condition.Status == status
-}
@@ -38,22 +38,22 @@ const (
 )

 // Snapshot contains persisted observations only, without credentials or live evidence.
-// Failure detail mapping will be added with capability assessment, not intent transitions.
 type Snapshot struct {
 	Phase            Phase
 	ObservedRevision int64
 	Readiness        Readiness
 	ReportedVersion  string
+	Failure          Failure
 }

 // Instance protects registration state and pure lifecycle transitions.
 // Reconstitution does not establish live capability evidence, even for a Ready snapshot.
-// This initial slice deliberately exposes no operation that authorizes provisioning.
 type Instance struct {
 	target     ObservationTarget
 	snapshot   Snapshot
 	deleting   bool
 	extensions ExtensionSupport
+	evidence   *CapabilityObservation
 }

 func Reconstitute(target ObservationTarget, snapshot Snapshot, deleting bool) (*Instance, error) {
@@ -85,6 +85,8 @@ func (i *Instance) BeginValidation() error {
 	i.snapshot.Phase = PhaseValidating
 	i.snapshot.Readiness = Unknown
 	i.extensions = ExtensionSupport{}
+	i.evidence = nil
+	i.snapshot.Failure = NoFailure
 	return nil
 }

@@ -100,6 +102,8 @@ func (i *Instance) BeginDeletion() error {
 	i.snapshot.Phase = PhaseDeleting
 	i.snapshot.Readiness = Unknown
 	i.extensions = ExtensionSupport{}
+	i.evidence = nil
+	i.snapshot.Failure = NoFailure
 	return nil
 }

@@ -29,10 +29,10 @@ func TestInstanceAcceptsExtensionObservationForCurrentTarget(t *testing.T) {
 	target := value.Target()
 	snapshot := value.Snapshot()

-	if err := value.ObserveExtensions(target, instance.ObserveExtensionSupport([]string{"pg_trgm"})); err != nil {
+	if err := value.ObserveExtensions(target, instance.ObserveExtensionSupport([]string{testTrigram})); err != nil {
 		t.Fatal(err)
 	}
-	if got := value.CheckExtensions(instance.NewExtensionSet([]string{"pg_trgm"})); got.Decision != instance.ExtensionsAccepted {
+	if got := value.CheckExtensions(instance.NewExtensionSet([]string{testTrigram})); got.Decision != instance.ExtensionsAccepted {
 		t.Fatalf("CheckExtensions() = %v, want accepted", got)
 	}
 	if value.Snapshot() != snapshot {
@@ -52,20 +52,20 @@ func TestInstanceRejectsExtensionObservationForDifferentTarget(t *testing.T) {
 		t.Fatal(err)
 	}

-	if err := value.ObserveExtensions(different, instance.ObserveExtensionSupport([]string{"pg_trgm"})); err == nil {
+	if err := value.ObserveExtensions(different, instance.ObserveExtensionSupport([]string{testTrigram})); err == nil {
 		t.Fatal("observation for a different target was accepted")
 	}
-	if got := value.CheckExtensions(instance.NewExtensionSet([]string{"pg_trgm"})); got.Decision != instance.ExtensionSupportUnobserved {
+	if got := value.CheckExtensions(instance.NewExtensionSet([]string{testTrigram})); got.Decision != instance.ExtensionSupportUnobserved {
 		t.Fatalf("rejected observation changed support: %v", got)
 	}
 }

 func TestInstanceClearsExtensionObservationAcrossLifecycleBoundaries(t *testing.T) {
-	requested := instance.NewExtensionSet([]string{"pg_trgm"})
+	requested := instance.NewExtensionSet([]string{testTrigram})

 	t.Run("validation", func(t *testing.T) {
 		value := lifecycleInstance(t, instance.Snapshot{Phase: instance.PhaseReady}, false)
-		if err := value.ObserveExtensions(value.Target(), instance.ObserveExtensionSupport([]string{"pg_trgm"})); err != nil {
+		if err := value.ObserveExtensions(value.Target(), instance.ObserveExtensionSupport([]string{testTrigram})); err != nil {
 			t.Fatal(err)
 		}
 		if err := value.BeginValidation(); err != nil {
@@ -78,7 +78,7 @@ func TestInstanceClearsExtensionObservationAcrossLifecycleBoundaries(t *testing.

 	t.Run("deletion", func(t *testing.T) {
 		value := lifecycleInstance(t, instance.Snapshot{Phase: instance.PhaseReady}, true)
-		if err := value.ObserveExtensions(value.Target(), instance.ObserveExtensionSupport([]string{"pg_trgm"})); err == nil {
+		if err := value.ObserveExtensions(value.Target(), instance.ObserveExtensionSupport([]string{testTrigram})); err == nil {
 			t.Fatal("deleting instance accepted a new observation")
 		}
 		if err := value.BeginDeletion(); err != nil {
@@ -92,8 +92,8 @@ func TestInstanceClearsExtensionObservationAcrossLifecycleBoundaries(t *testing.

 func TestInstanceCanExplicitlyInvalidateExtensionObservation(t *testing.T) {
 	value := lifecycleInstance(t, instance.Snapshot{Phase: instance.PhaseValidating}, false)
-	requested := instance.NewExtensionSet([]string{"pg_trgm"})
-	if err := value.ObserveExtensions(value.Target(), instance.ObserveExtensionSupport([]string{"pg_trgm"})); err != nil {
+	requested := instance.NewExtensionSet([]string{testTrigram})
+	if err := value.ObserveExtensions(value.Target(), instance.ObserveExtensionSupport([]string{testTrigram})); err != nil {
 		t.Fatal(err)
 	}
 	if err := value.ObserveExtensions(value.Target(), instance.ExtensionSupport{}); err != nil {
@@ -0,0 +1,263 @@
+/*
+Copyright 2026.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package instance
+
+import "errors"
+
+// Failure 只表示安全类别；驱动错误、凭据和 Condition 文案留在应用边界。
+type Failure uint8
+
+const (
+	NoFailure Failure = iota
+	ObservationIncomplete
+	DependencyUnavailable
+	AuthenticationFailed
+	InsufficientPrivileges
+	RegistryIncompatible
+	RegistryNotUsable
+)
+
+// CheckResult 的零值表示未观察，不能视为成功。
+type CheckResult uint8
+
+const (
+	CheckUnobserved CheckResult = iota
+	CheckPassed
+	CheckUnavailable
+	CheckAuthenticationFailed
+	CheckInsufficientPrivileges
+)
+
+// ManagementChecks 分别记录所需能力；SQL 探测和同轮次关联由 adapter/application 保证。
+// Extensions 不代表任意扩展均可安装；具体请求仍需支持检查、执行及回读。
+type ManagementChecks struct {
+	Connection CheckResult
+	Metadata   CheckResult
+	Roles      CheckResult
+	Databases  CheckResult
+	Grants     CheckResult
+	Extensions CheckResult
+}
+
+func (c ManagementChecks) failure() Failure {
+	for _, check := range []CheckResult{c.Connection, c.Metadata, c.Roles, c.Databases, c.Grants, c.Extensions} {
+		switch check {
+		case CheckPassed:
+		case CheckUnavailable:
+			return DependencyUnavailable
+		case CheckAuthenticationFailed:
+			return AuthenticationFailed
+		case CheckInsufficientPrivileges:
+			return InsufficientPrivileges
+		default:
+			return ObservationIncomplete
+		}
+	}
+	return NoFailure
+}
+
+type RegistryState uint8
+
+const (
+	RegistryUnobserved RegistryState = iota
+	RegistryAbsent
+	RegistryNeedsMigration
+	RegistryUsable
+	RegistryUnsupported
+	RegistryUnavailable
+)
+
+// CapabilityObservation 是值对象，不包含连接、凭据或可变集合。
+type CapabilityObservation struct {
+	target   ObservationTarget
+	version  string
+	checks   ManagementChecks
+	registry RegistryState
+}
+
+func NewCapabilityObservation(target ObservationTarget, version string,
+	checks ManagementChecks, registry RegistryState,
+) (CapabilityObservation, error) {
+	if err := target.Validate(); err != nil {
+		return CapabilityObservation{}, err
+	}
+	return CapabilityObservation{target: target, version: version, checks: checks, registry: registry}, nil
+}
+
+func (o CapabilityObservation) managementFailure() Failure {
+	if failure := o.checks.failure(); failure != NoFailure {
+		return failure
+	}
+	if o.version == "" {
+		return ObservationIncomplete
+	}
+	return NoFailure
+}
+
+func (o CapabilityObservation) registryFailure() Failure {
+	switch o.registry {
+	case RegistryUsable:
+		return NoFailure
+	case RegistryAbsent, RegistryNeedsMigration:
+		return RegistryNotUsable
+	case RegistryUnsupported:
+		return RegistryIncompatible
+	case RegistryUnavailable:
+		return DependencyUnavailable
+	default:
+		return ObservationIncomplete
+	}
+}
+
+type PreparationDecision uint8
+
+const (
+	PreparationDenied PreparationDecision = iota
+	PreparationAllowed
+	AlreadyUsable
+)
+
+func (i *Instance) acceptObservation(o CapabilityObservation, phase Phase) error {
+	if !i.target.Matches(o.target) {
+		return errors.New("capability observation target does not match instance")
+	}
+	if i.deleting || i.snapshot.Phase != phase {
+		return errors.New("capability observation is not allowed in current lifecycle")
+	}
+	return nil
+}
+
+func (i *Instance) fail(failure Failure) {
+	i.evidence = nil
+	i.extensions = ExtensionSupport{}
+	i.snapshot.Readiness = NotReady
+	i.snapshot.Failure = failure
+	i.snapshot.ObservedRevision = i.target.Revision().Value()
+}
+
+// AssessManagement 只推进意图，不执行 registry 写入，也不完成 observedRevision。
+func (i *Instance) AssessManagement(o CapabilityObservation) error {
+	if err := i.acceptObservation(o, PhaseValidating); err != nil {
+		return err
+	}
+	if failure := o.managementFailure(); failure != NoFailure {
+		i.fail(failure)
+		return nil
+	}
+	if failure := o.registryFailure(); failure != NoFailure && failure != RegistryNotUsable {
+		i.fail(failure)
+		return nil
+	}
+	i.snapshot.Phase = PhaseInitializingRegistry
+	i.snapshot.Readiness = Unknown
+	i.snapshot.Failure = NoFailure
+	i.evidence = nil
+	return nil
+}
+
+// PlanRegistryPreparation 不证明 checkpoint 已落盘；应用层必须先保存意图再执行写入。
+func (i *Instance) PlanRegistryPreparation(o CapabilityObservation) (PreparationDecision, error) {
+	if err := i.acceptObservation(o, PhaseInitializingRegistry); err != nil {
+		return PreparationDenied, err
+	}
+	if failure := o.managementFailure(); failure != NoFailure {
+		i.fail(failure)
+		return PreparationDenied, nil
+	}
+	switch o.registry {
+	case RegistryUsable:
+		return AlreadyUsable, nil
+	case RegistryAbsent, RegistryNeedsMigration:
+		return PreparationAllowed, nil
+	default:
+		i.fail(o.registryFailure())
+		return PreparationDenied, nil
+	}
+}
+
+// RegistryPreparationResult 只能是安全失败或完整回读，不能表达裸操作成功。
+type RegistryPreparationResult struct {
+	observation CapabilityObservation
+	failure     Failure
+}
+
+func RegistryReadBack(o CapabilityObservation) RegistryPreparationResult {
+	return RegistryPreparationResult{observation: o}
+}
+
+func RegistryPreparationFailed(target ObservationTarget, failure Failure) (RegistryPreparationResult, error) {
+	if err := target.Validate(); err != nil {
+		return RegistryPreparationResult{}, err
+	}
+	if failure < ObservationIncomplete || failure > RegistryNotUsable {
+		return RegistryPreparationResult{}, errors.New("registry preparation requires a known failure category")
+	}
+	return RegistryPreparationResult{observation: CapabilityObservation{target: target}, failure: failure}, nil
+}
+
+func (i *Instance) AssessRegistryResult(result RegistryPreparationResult) error {
+	o := result.observation
+	if err := i.acceptObservation(o, PhaseInitializingRegistry); err != nil {
+		return err
+	}
+	if result.failure != NoFailure {
+		i.fail(result.failure)
+		return nil
+	}
+	i.assessComplete(o)
+	return nil
+}
+
+func (i *Instance) assessComplete(o CapabilityObservation) {
+	if failure := o.managementFailure(); failure != NoFailure {
+		i.fail(failure)
+		return
+	}
+	if failure := o.registryFailure(); failure != NoFailure {
+		i.fail(failure)
+		return
+	}
+	i.snapshot = Snapshot{Phase: PhaseReady, ObservedRevision: i.target.Revision().Value(),
+		Readiness: Ready, ReportedVersion: o.version}
+	i.evidence = &o
+}
+
+// AssessReadiness 每轮接收完整事实，失败立即撤销本轮供应能力。
+func (i *Instance) AssessReadiness(o CapabilityObservation) error {
+	if err := i.acceptObservation(o, PhaseReady); err != nil {
+		return err
+	}
+	if i.snapshot.ObservedRevision != i.target.Revision().Value() {
+		return i.BeginValidation()
+	}
+	if o.managementFailure() != NoFailure || o.registry == RegistryUnavailable {
+		i.snapshot.Phase = PhaseValidating
+	} else if o.registryFailure() != NoFailure {
+		i.snapshot.Phase = PhaseInitializingRegistry
+	}
+	i.assessComplete(o)
+	return nil
+}
+
+// RequireProvisioningReady 仅检查 Instance 前置条件，不授予 Tenant 所有权或外部写入许可。
+func (i *Instance) RequireProvisioningReady() error {
+	if i.deleting || i.evidence == nil || i.snapshot.Phase != PhaseReady || i.snapshot.Readiness != Ready ||
+		i.snapshot.ObservedRevision != i.target.Revision().Value() {
+		return errors.New("instance is not ready for provisioning")
+	}
+	return nil
+}
@@ -0,0 +1,352 @@
+/*
+Copyright 2026.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package instance_test
+
+import (
+	"testing"
+
+	"git.ddupan.top/panxiao81/ayatori/internal/database/domain/instance"
+)
+
+const testServerVersion = "17.6"
+
+func completeChecks() instance.ManagementChecks {
+	return instance.ManagementChecks{
+		Connection: instance.CheckPassed, Metadata: instance.CheckPassed,
+		Roles: instance.CheckPassed, Databases: instance.CheckPassed,
+		Grants: instance.CheckPassed, Extensions: instance.CheckPassed,
+	}
+}
+
+func capability(t *testing.T, value *instance.Instance, checks instance.ManagementChecks,
+	registry instance.RegistryState,
+) instance.CapabilityObservation {
+	t.Helper()
+	o, err := instance.NewCapabilityObservation(value.Target(), testServerVersion, checks, registry)
+	if err != nil {
+		t.Fatal(err)
+	}
+	return o
+}
+
+func readyInstance(t *testing.T) *instance.Instance {
+	t.Helper()
+	i := lifecycleInstance(t, instance.Snapshot{Phase: instance.PhaseInitializingRegistry}, false)
+	if err := i.AssessRegistryResult(instance.RegistryReadBack(capability(t, i, completeChecks(), instance.RegistryUsable))); err != nil {
+		t.Fatal(err)
+	}
+	if err := i.RequireProvisioningReady(); err != nil {
+		t.Fatal(err)
+	}
+	return i
+}
+
+func TestReadinessRequiresCompleteReadBack(t *testing.T) {
+	i := lifecycleInstance(t, instance.Snapshot{}, false)
+	if err := i.BeginValidation(); err != nil {
+		t.Fatal(err)
+	}
+	absent := capability(t, i, completeChecks(), instance.RegistryAbsent)
+	if err := i.AssessManagement(absent); err != nil {
+		t.Fatal(err)
+	}
+	if s := i.Snapshot(); s.Phase != instance.PhaseInitializingRegistry || s.ObservedRevision != 0 || s.Readiness != instance.Unknown {
+		t.Fatalf("management observation prematurely concluded readiness: %+v", s)
+	}
+	for range 2 {
+		decision, err := i.PlanRegistryPreparation(absent)
+		if err != nil || decision != instance.PreparationAllowed {
+			t.Fatalf("preparation: %v, %v", decision, err)
+		}
+		if i.RequireProvisioningReady() == nil {
+			t.Fatal("preparation authorized provisioning")
+		}
+	}
+	if err := i.AssessRegistryResult(instance.RegistryReadBack(absent)); err != nil {
+		t.Fatal(err)
+	}
+	if i.Snapshot().Failure != instance.RegistryNotUsable || i.RequireProvisioningReady() == nil {
+		t.Fatal("absent registry accepted as ready")
+	}
+	usable := capability(t, i, completeChecks(), instance.RegistryUsable)
+	decision, err := i.PlanRegistryPreparation(usable)
+	if err != nil || decision != instance.AlreadyUsable {
+		t.Fatalf("retry after external preparation: %v, %v", decision, err)
+	}
+	if err := i.AssessRegistryResult(instance.RegistryReadBack(usable)); err != nil {
+		t.Fatal(err)
+	}
+	if s := i.Snapshot(); s.Readiness != instance.Ready || s.ReportedVersion != testServerVersion || s.ObservedRevision != i.Target().Revision().Value() {
+		t.Fatalf("complete observation not accepted: %+v", s)
+	}
+	if err := i.RequireProvisioningReady(); err != nil {
+		t.Fatal(err)
+	}
+}
+
+// 重启只恢复 checkpoint；依赖稍后恢复时必须重新取得完整事实。
+func TestReadinessRecoveryAndInvalidation(t *testing.T) {
+	i := readyInstance(t)
+	restored, err := instance.Reconstitute(i.Target(), i.Snapshot(), false)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if restored.RequireProvisioningReady() == nil {
+		t.Fatal("persisted Ready fabricated fresh evidence")
+	}
+	for range 2 {
+		if err := restored.AssessReadiness(capability(t, restored, completeChecks(), instance.RegistryUsable)); err != nil {
+			t.Fatal(err)
+		}
+		if err := restored.RequireProvisioningReady(); err != nil {
+			t.Fatal(err)
+		}
+	}
+	if err := restored.BeginValidation(); err != nil {
+		t.Fatal(err)
+	}
+	if restored.RequireProvisioningReady() == nil {
+		t.Fatal("validation retained evidence")
+	}
+	deleted, err := instance.Reconstitute(i.Target(), i.Snapshot(), true)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if deleted.RequireProvisioningReady() == nil {
+		t.Fatal("deletion allowed provisioning")
+	}
+	if err := deleted.BeginDeletion(); err != nil {
+		t.Fatal(err)
+	}
+	if deleted.RequireProvisioningReady() == nil {
+		t.Fatal("deleting checkpoint allowed provisioning")
+	}
+}
+
+func TestEachManagementCheckIsRequired(t *testing.T) {
+	for field := range 6 {
+		for _, result := range []instance.CheckResult{instance.CheckUnobserved, instance.CheckUnavailable,
+			instance.CheckAuthenticationFailed, instance.CheckInsufficientPrivileges, 255} {
+			checks := completeChecks()
+			fields := []*instance.CheckResult{&checks.Connection, &checks.Metadata, &checks.Roles,
+				&checks.Databases, &checks.Grants, &checks.Extensions}
+			*fields[field] = result
+			i := readyInstance(t)
+			if err := i.AssessReadiness(capability(t, i, checks, instance.RegistryUsable)); err != nil {
+				t.Fatal(err)
+			}
+			if s := i.Snapshot(); s.Phase != instance.PhaseValidating || s.Readiness != instance.NotReady ||
+				s.Failure == instance.NoFailure || i.RequireProvisioningReady() == nil {
+				t.Fatalf("check %d result %d accepted: %+v", field, result, s)
+			}
+		}
+	}
+}
+
+func TestRegistryDecisionsAndReadinessLoss(t *testing.T) {
+	for _, tc := range []struct {
+		state    instance.RegistryState
+		decision instance.PreparationDecision
+		failure  instance.Failure
+	}{
+		{instance.RegistryUsable, instance.AlreadyUsable, instance.NoFailure},
+		{instance.RegistryAbsent, instance.PreparationAllowed, instance.RegistryNotUsable},
+		{instance.RegistryNeedsMigration, instance.PreparationAllowed, instance.RegistryNotUsable},
+		{instance.RegistryUnsupported, instance.PreparationDenied, instance.RegistryIncompatible},
+		{instance.RegistryUnavailable, instance.PreparationDenied, instance.DependencyUnavailable},
+		{instance.RegistryUnobserved, instance.PreparationDenied, instance.ObservationIncomplete},
+		{255, instance.PreparationDenied, instance.ObservationIncomplete},
+	} {
+		i := lifecycleInstance(t, instance.Snapshot{Phase: instance.PhaseInitializingRegistry}, false)
+		o := capability(t, i, completeChecks(), tc.state)
+		decision, err := i.PlanRegistryPreparation(o)
+		if err != nil || decision != tc.decision {
+			t.Fatalf("registry %d: %v, %v", tc.state, decision, err)
+		}
+		i = readyInstance(t)
+		if err := i.AssessReadiness(o); err != nil {
+			t.Fatal(err)
+		}
+		if i.Snapshot().Failure != tc.failure {
+			t.Fatalf("registry %d: %+v", tc.state, i.Snapshot())
+		}
+		if tc.state != instance.RegistryUsable {
+			wantPhase := instance.PhaseInitializingRegistry
+			if tc.state == instance.RegistryUnavailable {
+				wantPhase = instance.PhaseValidating
+			}
+			if i.Snapshot().Phase != wantPhase || i.RequireProvisioningReady() == nil {
+				t.Fatal("registry drift retained readiness")
+			}
+		}
+	}
+}
+
+func TestInitializationRejectsIncompleteOrFailedManagement(t *testing.T) {
+	for _, tc := range []struct {
+		checks   instance.ManagementChecks
+		registry instance.RegistryState
+		failure  instance.Failure
+	}{
+		{instance.ManagementChecks{}, instance.RegistryUsable, instance.ObservationIncomplete},
+		{completeChecks(), instance.RegistryUnsupported, instance.RegistryIncompatible},
+		{completeChecks(), instance.RegistryUnavailable, instance.DependencyUnavailable},
+	} {
+		i := lifecycleInstance(t, instance.Snapshot{Phase: instance.PhaseValidating}, false)
+		if err := i.AssessManagement(capability(t, i, tc.checks, tc.registry)); err != nil {
+			t.Fatal(err)
+		}
+		if s := i.Snapshot(); s.Phase != instance.PhaseValidating || s.Failure != tc.failure ||
+			s.ObservedRevision != i.Target().Revision().Value() || s.Readiness != instance.NotReady {
+			t.Fatalf("invalid management accepted: %+v", s)
+		}
+	}
+	i := lifecycleInstance(t, instance.Snapshot{Phase: instance.PhaseInitializingRegistry}, false)
+	decision, err := i.PlanRegistryPreparation(capability(t, i, instance.ManagementChecks{}, instance.RegistryAbsent))
+	if err != nil || decision != instance.PreparationDenied || i.Snapshot().Failure != instance.ObservationIncomplete {
+		t.Fatal("incomplete management allowed registry writes")
+	}
+	if err := i.AssessRegistryResult(instance.RegistryReadBack(capability(t, i, completeChecks(), instance.RegistryUsable))); err != nil {
+		t.Fatal(err)
+	}
+	if err := i.RequireProvisioningReady(); err != nil {
+		t.Fatal("dependency recovery did not restore readiness", err)
+	}
+}
+
+func TestReadinessMethodsRejectWrongPhaseAndDeletion(t *testing.T) {
+	for _, deleting := range []bool{false, true} {
+		for _, phase := range []instance.Phase{instance.PhasePending, instance.PhaseValidating,
+			instance.PhaseInitializingRegistry, instance.PhaseReady, instance.PhaseDeleting} {
+			for _, operation := range []struct {
+				phase instance.Phase
+				apply func(*instance.Instance, instance.CapabilityObservation) error
+			}{
+				{instance.PhaseValidating, (*instance.Instance).AssessManagement},
+				{instance.PhaseReady, (*instance.Instance).AssessReadiness},
+				{instance.PhaseInitializingRegistry, func(i *instance.Instance, o instance.CapabilityObservation) error {
+					_, err := i.PlanRegistryPreparation(o)
+					return err
+				}},
+				{instance.PhaseInitializingRegistry, func(i *instance.Instance, o instance.CapabilityObservation) error {
+					return i.AssessRegistryResult(instance.RegistryReadBack(o))
+				}},
+			} {
+				if !deleting && operation.phase == phase {
+					continue
+				}
+				i := lifecycleInstance(t, instance.Snapshot{Phase: phase}, deleting)
+				before := i.Snapshot()
+				if err := operation.apply(i, capability(t, i, completeChecks(), instance.RegistryUsable)); err == nil {
+					t.Fatalf("phase %s deleting=%t accepted operation for %s", phase, deleting, operation.phase)
+				}
+				if i.Snapshot() != before {
+					t.Fatal("rejected operation mutated snapshot")
+				}
+			}
+		}
+	}
+}
+
+func TestOldGenerationObservationDoesNotReplaceEvidence(t *testing.T) {
+	i := readyInstance(t)
+	target := i.Target()
+	revision, err := instance.NewRevision(target.Revision().Value() + 1)
+	if err != nil {
+		t.Fatal(err)
+	}
+	other, err := instance.NewObservationTarget(target.Identity(), revision, target.Definition())
+	if err != nil {
+		t.Fatal(err)
+	}
+	o, err := instance.NewCapabilityObservation(other, testServerVersion, completeChecks(), instance.RegistryUsable)
+	if err != nil {
+		t.Fatal(err)
+	}
+	before := i.Snapshot()
+	if err := i.AssessReadiness(o); err == nil || i.Snapshot() != before {
+		t.Fatal("mismatched generation observation was accepted")
+	}
+	if err := i.RequireProvisioningReady(); err != nil {
+		t.Fatal("rejected unrelated input changed previously accepted evidence", err)
+	}
+}
+
+func TestPreparationFailureCannotEstablishReadiness(t *testing.T) {
+	i := lifecycleInstance(t, instance.Snapshot{Phase: instance.PhaseInitializingRegistry}, false)
+	for _, failure := range []instance.Failure{instance.DependencyUnavailable, instance.AuthenticationFailed,
+		instance.InsufficientPrivileges, instance.RegistryIncompatible} {
+		result, err := instance.RegistryPreparationFailed(i.Target(), failure)
+		if err != nil {
+			t.Fatal(err)
+		}
+		if err := i.AssessRegistryResult(result); err != nil {
+			t.Fatal(err)
+		}
+		if s := i.Snapshot(); s.Failure != failure || s.Readiness != instance.NotReady ||
+			s.Phase != instance.PhaseInitializingRegistry || i.RequireProvisioningReady() == nil {
+			t.Fatalf("failed operation accepted: %+v", s)
+		}
+	}
+	for _, failure := range []instance.Failure{instance.NoFailure, 255} {
+		if _, err := instance.RegistryPreparationFailed(i.Target(), failure); err == nil {
+			t.Fatal("invalid failure accepted")
+		}
+	}
+}
+
+func TestCapabilityInputsAndLifecycleGuards(t *testing.T) {
+	i := readyInstance(t)
+	if _, err := instance.NewCapabilityObservation(instance.ObservationTarget{}, testServerVersion,
+		completeChecks(), instance.RegistryUsable); err == nil {
+		t.Fatal("invalid target accepted")
+	}
+	if _, err := instance.RegistryPreparationFailed(instance.ObservationTarget{}, instance.DependencyUnavailable); err == nil {
+		t.Fatal("invalid failure target accepted")
+	}
+	for _, method := range []func(instance.CapabilityObservation) error{
+		i.AssessManagement, i.AssessReadiness,
+		func(o instance.CapabilityObservation) error { _, err := i.PlanRegistryPreparation(o); return err },
+		func(o instance.CapabilityObservation) error {
+			return i.AssessRegistryResult(instance.RegistryReadBack(o))
+		},
+	} {
+		before := i.Snapshot()
+		if err := method(instance.CapabilityObservation{}); err == nil || i.Snapshot() != before {
+			t.Fatal("mismatched observation accepted or mutated state")
+		}
+	}
+	old := i.Snapshot()
+	old.ObservedRevision = 0
+	changed := lifecycleInstance(t, old, false)
+	if err := changed.AssessReadiness(capability(t, changed, completeChecks(), instance.RegistryUsable)); err != nil {
+		t.Fatal(err)
+	}
+	if s := changed.Snapshot(); s.Phase != instance.PhaseValidating || s.ObservedRevision != 0 || s.Readiness != instance.Unknown {
+		t.Fatalf("changed generation accepted old checkpoint: %+v", s)
+	}
+	o, err := instance.NewCapabilityObservation(i.Target(), "", completeChecks(), instance.RegistryUsable)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := i.AssessReadiness(o); err != nil {
+		t.Fatal(err)
+	}
+	if i.Snapshot().Failure != instance.ObservationIncomplete {
+		t.Fatal("missing version accepted")
+	}
+}
Author	SHA1	Message	Date
panxiao81	cd0d3a70ae	feat: 实现 Instance Ready 领域判定与恢复规则 Verify / test (pull_request) Successful in 7m20s Details Verify / lint (pull_request) Successful in 7m51s Details	2026-09-21 07:49:41 +00:00
panxiao81	984c0aee73	Merge pull request 'feat: 迁移 Database Instance 领域基线与扩展观测' (#4 ) from feat/database-instance-domain into main Verify / test (push) Successful in 6m0s Details Verify / lint (push) Successful in 6m23s Details Reviewed-on: #4	2026-09-21 07:42:08 +00:00
panxiao81	813b4341d1	test: 复用扩展观测测试常量 Verify / test (pull_request) Successful in 5m43s Details Verify / lint (pull_request) Successful in 6m3s Details	2026-09-21 07:33:10 +00:00
panxiao81	5a7b38ad26	feat: 接入 Instance 扩展能力观测 Verify / test (pull_request) Successful in 4m16s Details Verify / lint (pull_request) Failing after 4m10s Details	2026-09-21 06:32:40 +00:00
panxiao81	8ac6283573	feat: 迁移 Database Instance 领域基线	2026-09-21 06:32:40 +00:00
panxiao81	33fb3ec972	Merge pull request 'docs: 明确 Ayatori 控制面与产品范围' (#3 ) from docs/ayatori-architecture into main Verify / test (push) Successful in 4m39s Details Verify / lint (push) Successful in 3m47s Details	2026-09-21 06:32:19 +00:00
panxiao81	86953cd91a	docs: 对齐产品里程碑与集成验证要求 Verify / test (pull_request) Successful in 3m14s Details Verify / lint (pull_request) Successful in 3m45s Details	2026-09-21 05:21:22 +00:00
panxiao81	ddd717209e	chore: 添加 homelab 知识维护 skill Verify / test (pull_request) Failing after 36s Details Verify / lint (pull_request) Failing after 39s Details	2026-09-21 05:14:04 +00:00
panxiao81	2756803ba4	docs: 保留 DBaaS 已批准设计合同 Verify / test (pull_request) Successful in 10m35s Details Verify / lint (pull_request) Successful in 12m9s Details	2026-09-20 20:38:53 +00:00
panxiao81	76752c8443	docs: 统一 Database API 到 Ayatori 域	2026-09-20 20:38:53 +00:00
panxiao81	36138f835a	docs: 允许 Database 模块无兼容负担重构	2026-09-20 20:38:52 +00:00
panxiao81	de8b7f9b04	docs: 记录 Compute 方向与 Database 模块合并	2026-09-20 20:38:51 +00:00
panxiao81	e6b9980b3e	docs: 解耦内置 API 与上游实现组件	2026-09-20 20:38:50 +00:00
panxiao81	641db531a8	docs: 按实际管理缺口限定产品范围	2026-09-20 20:38:49 +00:00
panxiao81	7b841d7dba	docs: 明确 API machinery 与领域控制循环边界	2026-09-20 20:38:48 +00:00