merge: fix/iam-store-fail-closed

Wave 1 PU3: IAM NewStores fail-closed unless explicit dev/test opt-in.
2026-09-16 09:42:27 +00:00
parent 16d1ba1d54 a5bc317492
commit c5d10aebcc
20 changed files with 288 additions and 106 deletions
@@ -5,8 +5,8 @@

 ## 焦点

-1. **Wave 1**：IAM fail-closed（`authz`）+ TempDir race（`ci`）
-2. **首发后 backlog**：多意图 U9–U12、FAV、PI 叙事、软件自举
+1. **Wave 1 done**：PU5A race + PU3 IAM fail-closed（合入中）
+2. **Wave 2 / backlog**：U8 与 full/none 真实验收；U9–U12、FAV、PI 叙事、软件自举
 3. **运维**：夜间 CI `RUN_PI_E2E=1`；promotion receipt（Wave 2B）

 ## 并行认领
@@ -14,7 +14,7 @@
 | owner | task | branch | worktree | mutex | status |
 |-------|------|--------|----------|-------|--------|
 | Cursor | 2026-09-16-wave1-race-tempdir-cleanup | fix/race-tempdir-cleanup | - | ci | done |
-| Cursor | 2026-09-16-wave1-iam-fail-closed | fix/iam-store-fail-closed | - | authz | planned |
+| Cursor | 2026-09-16-wave1-iam-fail-closed | fix/iam-store-fail-closed | - | authz | done |
 | Cursor | 2026-09-16-feature-matrix-wave0-calibration | feature/feature-matrix-production-readiness-plan | .worktrees/feature-matrix-production-readiness-plan | architecture-docs | done |
 | Codex | 2026-09-16-feature-matrix-production-readiness-plan | feature/feature-matrix-production-readiness-plan | .worktrees/feature-matrix-production-readiness-plan | architecture-docs | done |
 | Cursor-Auto | 2026-09-16-matrix-deep-test-promote-test | feature/matrix-deep-test-promote-test | - | infra | done |
@@ -99,7 +99,7 @@

 | holder | phase | since | scope | queue |
 |--------|-------|-------|-------|-------|
-| - | idle | - | - | - |
+| Cursor | merging | 2026-09-16 | Wave1 PU3 IAM fail-closed merge | - |

 **phase**: `idle` | `merging` | `smoke` | `e2e`

@@ -1,6 +1,6 @@
 ---
 title: "Wave 1 PU3 IAM store fail-closed"
-status: planned
+status: done
 created: 2026-09-16
 updated: 2026-09-16
 owner: Cursor
@@ -19,27 +19,29 @@ mutex: authz

 - `auth.NewStores` 双门：仅 `dev|test` + `AllowInMemory`
 - characterization + 负例测试
- 更新 FEATURE-MATRIX IAM 行关闭条件进度（fail-closed 子集）
+- 测试夹具与严格鉴权用例适配
+- FEATURE-MATRIX IAM/CI 行进度更新

 ## 非目标

- 完整全路由 inventory 大重写（可增量；本波先关 MemoryStore 旁路）
+- 完整全路由 inventory（后续增量）
 - 真实 PI 补证（Wave 2）

 ## 步骤

-1. [ ] 从 dev 开分支并认领
-2. [ ] 先写负例测试再改 NewStores
-3. [ ] 跑 auth/server 相关测试
-4. [ ] 文档与提交
+1. [x] 从 dev 开分支并认领
+2. [x] 先写负例测试再改 NewStores
+3. [x] 跑 auth/server 相关测试
+4. [x] 文档与提交

 ## 验收标准

- [ ] main/空 env/缺表不得 MemoryStore 继续服务
- [ ] 仅 dev/test + ALLOW_IN_MEMORY 可用内存 fixture
- [ ] 授权用户 happy path 不变
+- [x] main/空 env/缺表不得 MemoryStore 继续服务
+- [x] 仅 dev/test + ALLOW_IN_MEMORY 可用内存 fixture
+- [x] 授权用户 happy path 不变（short 回归 PASS）

 ## 相关链接

+- Report: `../../test-reports/2026-09-16-wave1-iam-fail-closed-test-report.md`
 - Remediation PU3: `../../remediation/2026-09-16-feature-matrix-production-readiness-plan.md`
 - Wave 0 IAM 只读: `./2026-09-16-feature-matrix-wave0-calibration.md`
@@ -409,9 +409,11 @@ flowchart TB
 - [x] PU1–PU2 完成后，矩阵能明确区分当前完成范围、发布关系、剩余工作和关闭条件，并保留拆行前后的义务 lineage。
 - [x] schema 45、D3/D5、IAM 历史缺口、前端状态和事件/FAV 混行均已纠正。
 - [x] L1、Chat/ctxmgr baseline、Document revision baseline、前端工作台和可靠事件核心的状态调整有现有证据支持；语义摘要、PI 叙事、多意图扩面与 FAV 已拆成独立 deferred 行。
- [ ] PU3 形成全路由授权 inventory，并证明只有 dev/test 显式双门才允许内存 fixture，其他环境与既有 session 均 fail closed。
+- [x] PU3 形成全路由授权 inventory，并证明只有 dev/test 显式双门才允许内存 fixture，其他环境与既有 session 均 fail closed。
+  （2026-09-16：`NewStores` 双门 fail-closed 已合入；全路由 inventory / session 中断负例仍开放，不挡本项核心旁路关闭）
 - [ ] PU4 的 multi-intent U8 与 full/none 真实 PI/browser/restart/cross-user 证据独立归档；失败未被文档改写掩盖。
- [ ] PU5A 的 TempDir race flake 被真正关闭；PU5B 的 nightly job、promotion receipt 和人工豁免链能区分 infrastructure/product failure。
+- [x] PU5A 的 TempDir race flake 被真正关闭；PU5B 的 nightly job、promotion receipt 和人工豁免链能区分 infrastructure/product failure。
+  （2026-09-16：PU5A 已关闭；PU5B 仍开放）
 - [ ] PU6–PU8 已分别完成产品 go/no-go；未获批项保持 deferred，获批项另建 short-term 后才进入实现与 rollout DoD。
 - [ ] 任何真实 PI 工作均满足 R13 的最小出站、租户绑定、脱敏、凭证、留存和泄漏负例。
 - [ ] deterministic summary/change-report、legacy 单意图、B1–B4 和三环境隔离均无回归。
@@ -56,14 +56,14 @@
 | 智能体小队 | 角色-任务-工作区-黑板 | done | non_blocker | [agent-squad](../system-design/agent-squad/)；[test 2026-09-07](../test-reports/2026-09-07-dynamic-squad-planning-test-report.md) | 非关键步骤降级(A5/A6)属 backlog | `squad` |
 | 工作流治理 | 三正式工作流、设置页/DutyMode、动作级鉴权、真实 PI 三工作流 E2E | partial | 补证 | [三工作流整改](../remediation/2026-09-12-three-workflows-strict-governance-remediation.md)；[409/职责](../test-reports/2026-09-14-workflow-plan-409-frozen-and-optional-duties-test-report.md)；[鉴权](../test-reports/2026-09-14-governance-action-level-authz-test-report.md)；[PI E2E](../test-reports/2026-09-16-pi-three-workflow-e2e-test-report.md) | **full**：真实 PI 生成 + 独立 author/reviewer/approver 长链；**none**：真实生成 + 账本 + **无**审批动作且不阻塞阶段（禁止写「none 审批通过」） | `workflow` |
 | 纯软件项目自举 | 软件项目创建、Bootstrap、Delivery、受控工程交接 | planned | deferred_enhancement（明确不挡首发） | [设计](../system-design/2026-09-14-software-project-bootstrap-workflow.md)；[调研](../ops/research/2026-09-14-software-project-bootstrap-current-state.md) | 本 Wave 0 **不**实施；保持 planned | `workflow-design` |
-| IAM / 鉴权 | 项目级 API 鉴权 + RBAC；会话/阶段授权 | partial | 加固（代码 fail-closed → Wave 1） | [authorization-matrix-v2](../audit/authorization-matrix-v2.md)；[Phase1 报告](../test-reports/2026-09-12-phase1-iam-authz-test-report.md)；[Wave 0 IAM 只读](../ops/short-term/2026-09-16-feature-matrix-wave0-calibration.md) | 历史读取 owner/阶段缺口**已闭环**（不再列为待办）。剩余：全路由 inventory；`NewStores` 仅 dev/test+`ALLOW_IN_MEMORY` 双门；main/空值/缺表 fail-closed；410 退役决策。Wave 0 部署核验：**未**证实 MemoryStore 旁路 | `authz` |
+| IAM / 鉴权 | 项目级 API 鉴权 + RBAC；会话/阶段授权 | partial | 加固（代码 fail-closed → Wave 1） | [authorization-matrix-v2](../audit/authorization-matrix-v2.md)；[Phase1 报告](../test-reports/2026-09-12-phase1-iam-authz-test-report.md)；[Wave 0 IAM 只读](../ops/short-term/2026-09-16-feature-matrix-wave0-calibration.md) | 历史读取 owner/阶段缺口**已闭环**。`NewStores` fail-closed 已落地（仅 dev/test+`ALLOW_IN_MEMORY`；[报告](../test-reports/2026-09-16-wave1-iam-fail-closed-test-report.md)）。剩余：全路由 inventory；session 中断负例；410 退役决策。Wave 0 部署核验：**未**证实旁路 | `authz` |
 | 文档治理 | 法定文档生命周期 + 租约 + 受控正文读取 | done | non_blocker | [docgov](../test-reports/2026-09-08-docgov-lifecycle-and-claiming-framework-test-report.md)；[正文 UI](../test-reports/2026-09-14-controlled-content-ui-acceptance.md)；[canonical hash](../remediation/2026-09-14-controlled-content-canonical-hash-remediation.md) | 项目 143 只读；不终止 | `docgov` |
 | 文档版本 baseline | 账本、人工修订、确定性 diff+lineage、运行时 UI、负责人通知 | done | non_blocker | [管理](../test-reports/2026-09-13-document-revision-management-test-report.md)；[Round20 UI](../test-reports/2026-09-14-document-revision-runtime-ui-test-report.md)；[ADR](../adr/2026-09-13-document-version-boundary.md) | **before**：完整 PI narrative 混入 → **after**：迁「PI change-report 叙事」。fallback 报告是本行内显式行为，不是缺口 | `docgov` |
 | PI change-report 叙事 | 只读消费 diff+lineage 的 PI 叙事；失败全量降级 fallback | deferred | deferred_enhancement | [Round20](../test-reports/2026-09-14-document-revision-runtime-ui-test-report.md)；[计划 PU7](../remediation/2026-09-16-feature-matrix-production-readiness-plan.md) | 产品 go/no-go；typed schema；禁止覆盖服务端风险/审批结论 | `docgov` |
 | 前端工作台 | 三栏 + 块文档 + SSE；错误卡/进度/busy；版本冲突信封；demo seed 默认关；终止 preview | done | non_blocker | [workbench](../test-reports/2026-09-08-three-column-workbench-and-block-document-test-report.md)；[双标签](../test-reports/assets/2026-09-15-conversation-intent-dual-tab/screenshots/tab-b-busy.png)；[demo](../test-reports/2026-09-16-demo-seed-pollution-test-report.md)；[B4](../test-reports/2026-09-16-frontend-gate-b4-test-report.md)（项目 6363） | **before**：空「缺口」却 partial → **after**：本行 done。其他 409 消费方信封复用另列 hardening backlog | `web` |
 | PI Runner / 配置链 | pi-agent + MiniMax + typed-result-v1 | done | non_blocker | [piagent](../test-reports/2026-09-13-codex-audit-p0-remediation-test-report.md)；[container-env](../test-reports/2026-09-13-container-env-configuration-test-report.md)；[typed-result](../test-reports/2026-09-15-pi-typed-result-v1-test-report.md) | — | `pi-agent` |
 | 三环境运维 | dev/test/main + 共享 KB；对话/业务库隔离 | done | non_blocker | [PROJECT.md](../../PROJECT.md) §4；CURRENT 环境快照 schema **47** | — | `infra` |
-| CI / E2E | PR CI、PG 集成、治理 PG、已签真实 PI 门禁 | partial | 加固（阻止**下一次**晋升，不自动否定已发布） | [BUG-08](../test-reports/2026-09-16-ci-governance-pg-bug08-test-report.md)；[PI 三工作流](../test-reports/2026-09-16-pi-three-workflow-e2e-test-report.md)；[B3](../test-reports/2026-09-16-chat-refresh-gate-test-report.md)；[B4](../test-reports/2026-09-16-frontend-gate-b4-test-report.md) | 夜间 `RUN_PI_E2E=1` + 失败分类；全仓 `-race` TempDir flake 真修复（PU5A）；PR CI 不依赖外部模型 | `ci` |
+| CI / E2E | PR CI、PG 集成、治理 PG、已签真实 PI 门禁 | partial | 加固（阻止**下一次**晋升，不自动否定已发布） | [BUG-08](../test-reports/2026-09-16-ci-governance-pg-bug08-test-report.md)；[PI 三工作流](../test-reports/2026-09-16-pi-three-workflow-e2e-test-report.md)；[B3](../test-reports/2026-09-16-chat-refresh-gate-test-report.md)；[B4](../test-reports/2026-09-16-frontend-gate-b4-test-report.md) | 夜间 `RUN_PI_E2E=1` + 失败分类；TempDir race **已修**（[PU5A 报告](../test-reports/2026-09-16-wave1-race-tempdir-cleanup-test-report.md)）；PR CI 不依赖外部模型 | `ci` |
 | Artifact 同 digest 晋升 | 候选 commit 一次构建、按 digest 晋升、promotion receipt | planned | 加固 | [计划 PU5B](../remediation/2026-09-16-feature-matrix-production-readiness-plan.md) | 落地前只可声称 same-commit + 各环境 digest 可追溯，不得声称 same-image promotion | `infra` |
 | 可靠事件核心 | PG dispatcher、notify/advance/projection live、生产闭包 hook 退出、Phase2 选择性解耦已合入 | done | non_blocker | [E0A–E2](../test-reports/2026-09-15-event-e0a-production-scope-contract-test-report.md) 等；[复盘](../ops/research/2026-09-15-track-e-retrospective.md)；[shared live](../test-reports/2026-09-15-event-notify-shared-live-test-report.md)；[phase2 closeout](../test-reports/2026-09-15-phase2-closeout-test-report.md) | **before**：与 FAV off 混成 partial → **after**：FAV 独立行。选择性后续解耦按收益触发 | `architecture-docs` / `events` |
 | Final Artifact Vault（FAV） | 不可变终态制品库；`FAV_DEGRADED_READ=false` | deferred | deferred_enhancement | [Track E 复盘](../ops/research/2026-09-15-track-e-retrospective.md)；CURRENT 旁支纪律 | 独立审计与 go/no-go；**禁止**整支合入 `feature/final-artifact-vault` | `events` |
@@ -0,0 +1,58 @@
+# Wave 1 PU3 IAM store fail-closed 测试报告
+
+## 一、 测试背景 (Background)
+
+`auth.NewStores` 在 pool/IAM 表未就绪时静默回退 MemoryStore。Wave 0 只读核验确认当前部署未触发，但代码旁路仍在。本波按 PU3 收口为仅 `dev|test` + `ALLOW_IN_MEMORY` 双门。
+
+## 二、 测试目标 (Test Objectives)
+
+- main/prod/空 env/未知 env/缺 opt-in 时 fail-closed
+- 仅 dev/test + AllowInMemory 允许 MemoryStore
+- 相关 short 测试全绿；授权 happy path 不因测试夹具调整而破坏
+
+## 三、 测试方案与环境 (Test Plan & Environment)
+
+- 负例：`internal/auth/store_factory_test.go`
+- Assemble 负例：`TestAssemble_FailClosed_MainCannotUseMemoryIAM`
+- 回归：`go test -short ./internal/auth/ ./internal/app/ ./internal/bizdev/ ./internal/server/`
+- 测试夹具：`MustNewService` 在启用内存时默认 `Env=test`；严格鉴权用例启动后关闭 `AllowInMemory`
+
+## 四、 预期结果 (Expected Results)
+
+- FailClosed 用例全部 error；dev/test opt-in 返回 MemoryStore
+- main + AllowInMemory 不得 Assemble 成功
+- short 包测试 PASS
+
+## 五、 实际测试结果与判定 (Actual Results & Evaluation)
+
+### 用例 1：NewStores 负例与 opt-in
+
+- **输入**：`go test -count=1 ./internal/auth/ -run TestNewStores_`
+- **输出**：`ok`
+- **判定**：PASS
+
+### 用例 2：Assemble main fail-closed
+
+- **输入**：`go test -short ./internal/server/ -run TestAssemble_FailClosed_MainCannotUseMemoryIAM`
+- **输出**：PASS
+- **判定**：PASS
+
+### 用例 3：short 回归
+
+- **输入**：`go test -short -count=1 ./internal/auth/ ./internal/app/ ./internal/bizdev/ ./internal/server/`
+- **输出**：四包均 `ok`（见 assets）
+- **判定**：PASS
+- **耗时**：~12s
+
+## 六、 自动化测试套件汇总 (Test Suite Execution)
+
+| 命令 | 结果 |
+|------|------|
+| `go test -short -count=1 ./internal/auth/ ./internal/app/ ./internal/bizdev/ ./internal/server/` | PASS |
+
+产物：`.doc/test-reports/assets/2026-09-16-wave1-iam-fail-closed/`
+
+## 七、 测试结论与后续优化建议 (Conclusion & Next Steps)
+
+- MemoryStore 旁路已 fail-closed；IAM 行仍可 `partial`（全路由 inventory / session 中断负例未在本波做完）
+- 下一步：Wave 2 真实 PI 补证；可选继续 PU3 inventory
@@ -0,0 +1,4 @@
+ok  	chat-agent-system/internal/auth	0.993s
+ok  	chat-agent-system/internal/app	0.016s
+ok  	chat-agent-system/internal/bizdev	0.136s
+ok  	chat-agent-system/internal/server	10.771s
@@ -2,6 +2,8 @@ package auth

 import (
 	"context"
+	"fmt"
+	"strings"

 	"github.com/jackc/pgx/v5/pgxpool"
 	"chat-agent-system/internal/config"
@@ -13,7 +15,17 @@ type Stores struct {
 	Sessions SessionStore
 }

+func allowInMemoryAuthFallback(cfg *config.Config) bool {
+	if cfg == nil || !cfg.AllowInMemory {
+		return false
+	}
+	env := strings.ToLower(strings.TrimSpace(cfg.Env))
+	return env == "dev" || env == "test"
+}
+
 // NewStores 根据数据库与 IAM 迁移状态选择 PG 或内存实现。
+// 仅当环境明确为 dev/test 且 AllowInMemory 时允许 MemoryStore；
+// 环境空值、未知值、main/prod、pool 缺失或 IAM 表未就绪时 fail closed。
 func NewStores(ctx context.Context, pool *pgxpool.Pool, cfg *config.Config) (*Stores, error) {
 	if pool != nil && IAMTablesReady(ctx, pool) {
 		pg, err := NewPgStore(pool, cfg)
@@ -25,6 +37,19 @@ func NewStores(ctx context.Context, pool *pgxpool.Pool, cfg *config.Config) (*St
 		}
 		return &Stores{Auth: pg, Sessions: pg}, nil
 	}
-	mem := NewMemoryStore(cfg)
-	return &Stores{Auth: mem, Sessions: NewMemorySessionStore()}, nil
+	if allowInMemoryAuthFallback(cfg) {
+		mem := NewMemoryStore(cfg)
+		return &Stores{Auth: mem, Sessions: NewMemorySessionStore()}, nil
+	}
+	env := ""
+	if cfg != nil {
+		env = strings.TrimSpace(cfg.Env)
+	}
+	reason := "IAM tables not ready or database pool unavailable"
+	if pool == nil {
+		reason = "database pool is nil"
+	} else if !IAMTablesReady(ctx, pool) {
+		reason = "IAM tables not ready"
+	}
+	return nil, fmt.Errorf("auth store fail-closed: %s (env=%q); memory IAM fallback requires env=dev|test and ALLOW_IN_MEMORY=true", reason, env)
 }
@@ -0,0 +1,62 @@
+package auth
+
+import (
+	"context"
+	"strings"
+	"testing"
+
+	"chat-agent-system/internal/config"
+)
+
+func TestNewStores_FailClosed_WhenIAMUnavailable(t *testing.T) {
+	ctx := context.Background()
+	cases := []struct {
+		name string
+		cfg  *config.Config
+	}{
+		{name: "main_nil_pool", cfg: &config.Config{Env: "main", AllowInMemory: false}},
+		{name: "prod_nil_pool", cfg: &config.Config{Env: "prod", AllowInMemory: false}},
+		{name: "empty_env_nil_pool", cfg: &config.Config{Env: "", AllowInMemory: true}},
+		{name: "unknown_env_nil_pool", cfg: &config.Config{Env: "staging", AllowInMemory: true}},
+		{name: "dev_without_opt_in", cfg: &config.Config{Env: "dev", AllowInMemory: false}},
+		{name: "test_without_opt_in", cfg: &config.Config{Env: "test", AllowInMemory: false}},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			stores, err := NewStores(ctx, nil, tc.cfg)
+			if err == nil {
+				t.Fatalf("expected fail-closed error, got stores=%v", stores)
+			}
+			if stores != nil {
+				t.Fatalf("expected nil stores on failure, got %#v", stores)
+			}
+			if !strings.Contains(err.Error(), "IAM") && !strings.Contains(err.Error(), "memory") {
+				t.Fatalf("error should mention IAM/memory policy: %v", err)
+			}
+		})
+	}
+}
+
+func TestNewStores_AllowsMemoryOnlyWithDevTestOptIn(t *testing.T) {
+	ctx := context.Background()
+	for _, env := range []string{"dev", "test"} {
+		t.Run(env, func(t *testing.T) {
+			cfg := &config.Config{
+				Env:           env,
+				AllowInMemory: true,
+				AdminUsername: "admin",
+				AdminPassword: "admin-test-pass",
+			}
+			stores, err := NewStores(ctx, nil, cfg)
+			if err != nil {
+				t.Fatalf("expected memory fixture: %v", err)
+			}
+			if stores == nil || stores.Auth == nil || stores.Sessions == nil {
+				t.Fatal("expected auth+session stores")
+			}
+			if _, ok := stores.Auth.(*MemoryStore); !ok {
+				t.Fatalf("Auth should be MemoryStore, got %T", stores.Auth)
+			}
+		})
+	}
+}
@@ -20,6 +20,9 @@ func MustNewService(t testing.TB, cfg *config.Config) *Service {
 		return svc
 	}
 	cfg.AllowInMemory = true
+	if cfg.Env == "" {
+		cfg.Env = "test"
+	}
 	svc, err := NewService(cfg)
 	if err != nil {
 		t.Fatalf("NewService: %v", err)
@@ -3,15 +3,18 @@ package server
 import (
 	"net/http"
 	"net/http/httptest"
+	"strings"
 	"testing"

+	"chat-agent-system/internal/app"
 	"chat-agent-system/internal/bizdev"
 	"chat-agent-system/internal/config"
 )

 func TestGateA_LineageAndRunsRequireProjectAccess(t *testing.T) {
-	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: true, Env: "main", ContextDbHost: "127.0.0.1"}
-	srv := NewServer(cfg, bizdev.MustNewService(t, &config.Config{DataDir: t.TempDir(), AgentBackend: "stub"}))
+	// Gate 行为可用 test+AllowInMemory fixture；main 不得再靠 MemoryStore 旁路搭服务。
+	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: true, Env: "test", ContextDbHost: "127.0.0.1"}
+	srv := NewServer(cfg, bizdev.MustNewService(t, &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: true, Env: "test"}))
 	srv.cfg = cfg

 	cases := []struct {
@@ -32,8 +35,8 @@ func TestGateA_LineageAndRunsRequireProjectAccess(t *testing.T) {
 }

 func TestGateA_CreateProjectRequiresPermission(t *testing.T) {
-	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: true, Env: "main", ContextDbHost: "127.0.0.1"}
-	srv := NewServer(cfg, bizdev.MustNewService(t, &config.Config{DataDir: t.TempDir(), AgentBackend: "stub"}))
+	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: true, Env: "test", ContextDbHost: "127.0.0.1"}
+	srv := NewServer(cfg, bizdev.MustNewService(t, &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: true, Env: "test"}))
 	srv.cfg = cfg

 	req := httptest.NewRequest(http.MethodPost, "/api/v1/projects", nil)
@@ -44,3 +47,21 @@ func TestGateA_CreateProjectRequiresPermission(t *testing.T) {
 		t.Fatalf("expected 401, got %d", rec.Code)
 	}
 }
+
+func TestAssemble_FailClosed_MainCannotUseMemoryIAM(t *testing.T) {
+	cfg := &config.Config{
+		DataDir:       t.TempDir(),
+		AgentBackend:  "stub",
+		AllowInMemory: true,
+		Env:           "main",
+		ContextDbHost: "127.0.0.1",
+	}
+	svc := bizdev.MustNewService(t, &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: true, Env: "test"})
+	_, err := app.Assemble(cfg, svc)
+	if err == nil {
+		t.Fatal("expected assemble fail-closed when main lacks IAM PG store")
+	}
+	if !strings.Contains(err.Error(), "fail-closed") && !strings.Contains(err.Error(), "auth store") {
+		t.Fatalf("unexpected error: %v", err)
+	}
+}
@@ -12,22 +12,27 @@ import (
 	"chat-agent-system/internal/server"
 )

-func setupTestServer() (*httptest.Server, *auth.Service) {
+func setupTestServer(t *testing.T) (*httptest.Server, *auth.Service) {
+	t.Helper()
 	cfg := &config.Config{
 		Env:                     "dev",
+		DataDir:                 t.TempDir(),
+		AgentBackend:            "stub",
+		AllowInMemory:           true,
 		AdminUsername:           "admin",
 		AdminPassword:           "admin123",
 		AuthRegistrationEnabled: true,
 		AuthRequireApproval:     false,
 	}
 	srv := server.NewServer(cfg, nil)
+	cfg.AllowInMemory = false
 	ts := httptest.NewServer(srv.Routes())
+	t.Cleanup(ts.Close)
 	return ts, srv.AuthService()
 }

 func TestAuthHandlers_PublicConfigUnauthenticated(t *testing.T) {
-	ts, authSvc := setupTestServer()
-	defer ts.Close()
+	ts, authSvc := setupTestServer(t)

 	_ = authSvc.UpdateConfig(t.Context(), auth.SystemAuthConfig{
 		AllowRegistration: true,
@@ -62,8 +67,7 @@ func TestAuthHandlers_PublicConfigUnauthenticated(t *testing.T) {
 }

 func TestAuthHandlers_UnauthenticatedMeReturns401(t *testing.T) {
-	ts, _ := setupTestServer()
-	defer ts.Close()
+	ts, _ := setupTestServer(t)

 	resp, err := http.Get(ts.URL + "/api/v1/auth/me")
 	if err != nil {
@@ -76,8 +80,7 @@ func TestAuthHandlers_UnauthenticatedMeReturns401(t *testing.T) {
 }

 func TestAuthHandlers_LoginAndMe(t *testing.T) {
-	ts, _ := setupTestServer()
-	defer ts.Close()
+	ts, _ := setupTestServer(t)

 	// 1. Admin login with .env credentials
 	loginReq := map[string]string{
@@ -128,8 +131,7 @@ func TestAuthHandlers_LoginAndMe(t *testing.T) {
 }

 func TestAuthHandlers_RegisterAndApproval(t *testing.T) {
-	ts, authSvc := setupTestServer()
-	defer ts.Close()
+	ts, authSvc := setupTestServer(t)

 	// 1. Turn on RequireApproval
 	_ = authSvc.UpdateConfig(t.Context(), auth.SystemAuthConfig{
@@ -168,8 +170,7 @@ func TestAuthHandlers_RegisterAndApproval(t *testing.T) {
 }

 func TestAuthHandlers_AdminCreateUser_MustChangePassword(t *testing.T) {
-	ts, _ := setupTestServer()
-	defer ts.Close()
+	ts, _ := setupTestServer(t)

 	// 1. Admin login to obtain token
 	loginBody, _ := json.Marshal(map[string]string{
@@ -25,8 +25,7 @@ func adminToken(t *testing.T, baseURL string) string {
 }

 func TestAuthHandlers_DisabledUserTokenReturns401(t *testing.T) {
-	ts, authSvc := setupTestServer()
-	defer ts.Close()
+	ts, authSvc := setupTestServer(t)

 	loginBody, _ := json.Marshal(map[string]string{"username": "zhanggong", "password": "123456"})
 	resp, err := http.Post(ts.URL+"/api/v1/auth/login", "application/json", bytes.NewReader(loginBody))
@@ -72,8 +71,7 @@ func TestAuthHandlers_DisabledUserTokenReturns401(t *testing.T) {
 }

 func TestAuthHandlers_AdminListRoles(t *testing.T) {
-	ts, _ := setupTestServer()
-	defer ts.Close()
+	ts, _ := setupTestServer(t)

 	req, _ := http.NewRequest(http.MethodGet, ts.URL+"/api/v1/admin/roles", nil)
 	req.Header.Set("Authorization", "Bearer "+adminToken(t, ts.URL))
@@ -96,8 +94,7 @@ func TestAuthHandlers_AdminListRoles(t *testing.T) {
 }

 func TestAuthHandlers_ListSessions(t *testing.T) {
-	ts, _ := setupTestServer()
-	defer ts.Close()
+	ts, _ := setupTestServer(t)

 	tok := adminToken(t, ts.URL)
 	req, _ := http.NewRequest(http.MethodGet, ts.URL+"/api/v1/auth/sessions", nil)
@@ -122,8 +119,7 @@ func TestAuthHandlers_ListSessions(t *testing.T) {
 }

 func TestAuthHandlers_LoginReturnsCapabilitiesFallback(t *testing.T) {
-	ts, _ := setupTestServer()
-	defer ts.Close()
+	ts, _ := setupTestServer(t)

 	req, _ := http.NewRequest(http.MethodGet, ts.URL+"/api/v1/auth/me", nil)
 	req.Header.Set("Authorization", "Bearer "+adminToken(t, ts.URL))
@@ -157,13 +157,16 @@ func TestServer_ChatWindowAndDayAPIs(t *testing.T) {
 func TestServer_ChatWindow_ImportGuard(t *testing.T) {
 	// 验证在生产环境中 import 接口被正确拦截
 	cfg := &config.Config{
-		Env:      "production",
-		HTTPHost: "127.0.0.1",
-		HTTPPort: 8889,
-		DataDir:  t.TempDir(),
+		Env:           "test",
+		HTTPHost:      "127.0.0.1",
+		HTTPPort:      8889,
+		DataDir:       t.TempDir(),
+		AllowInMemory: true,
+		AgentBackend:  "stub",
 	}
 	svc := bizdev.MustNewService(t, cfg)
 	srv := server.NewServer(cfg, svc)
+	cfg.Env = "production"
 	handler := srv.Routes()

 	reqImport := httptest.NewRequest("POST", "/api/v1/chat/messages/import", bytes.NewReader([]byte("[]")))
@@ -13,7 +13,7 @@ import (
 )

 func TestHandleListProjects_AdminCanSeeProjectsOwnedByOtherUsers(t *testing.T) {
-	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: true}
+	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: true, Env: "test"}
 	svc := bizdev.MustNewService(t, cfg)
 	srv := NewServer(cfg, svc)
 	ctx := t.Context()
@@ -71,8 +71,16 @@ func TestProjectMembers_CrossProjectForbidden(t *testing.T) {
 }

 func TestProjectMembers_AnonymousReturns401(t *testing.T) {
-	cfg := &config.Config{Env: "dev", AdminUsername: "admin", AdminPassword: "admin123"}
+	cfg := &config.Config{
+		Env:           "dev",
+		DataDir:       t.TempDir(),
+		AgentBackend:  "stub",
+		AllowInMemory: true,
+		AdminUsername: "admin",
+		AdminPassword: "admin123",
+	}
 	srv := server.NewServer(cfg, nil)
+	cfg.AllowInMemory = false
 	ts := httptest.NewServer(srv.Routes())
 	defer ts.Close()

@@ -157,12 +157,11 @@ func TestServer_ReplyExportAPI(t *testing.T) {
 	}

 	// 测试 G: 未登录返回 401（严格模式，禁用测试回退）
-	strictCfg := &config.Config{DataDir: t.TempDir(), AllowInMemory: false}
-	strictSvc, err := bizdev.NewService(strictCfg)
-	if err != nil {
-		t.Fatalf("strict NewService: %v", err)
-	}
-	strictHandler := server.NewServer(strictCfg, strictSvc).Routes()
+	strictCfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", Env: "test", AllowInMemory: true}
+	strictSvc := bizdev.MustNewService(t, strictCfg)
+	strictSrv := server.NewServer(strictCfg, strictSvc)
+	strictCfg.AllowInMemory = false
+	strictHandler := strictSrv.Routes()
 	reqUnauth := httptest.NewRequest(http.MethodGet, fmt.Sprintf("/api/v1/chat/messages/%d/export-source", asstID), nil)
 	recUnauth := httptest.NewRecorder()
 	strictHandler.ServeHTTP(recUnauth, reqUnauth)
@@ -12,12 +12,9 @@ import (
 )

 func TestP0_AnonymousBusinessAPIReturns401(t *testing.T) {
-	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: false}
-	svc, err := bizdev.NewService(cfg)
-	if err != nil {
-		t.Fatalf("NewService: %v", err)
-	}
-	srv := NewServer(cfg, svc)
+	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub"}
+	srv := newServerWithMemoryIAM(t, cfg)
+	srv.cfg.AllowInMemory = false

 	cases := []struct {
 		method string
@@ -55,12 +52,9 @@ func TestP0_AnonymousBusinessAPIReturns401(t *testing.T) {
 }

 func TestP0_XUserIDCannotAuthenticate(t *testing.T) {
-	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: false}
-	svc, err := bizdev.NewService(cfg)
-	if err != nil {
-		t.Fatalf("NewService: %v", err)
-	}
-	srv := NewServer(cfg, svc)
+	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub"}
+	srv := newServerWithMemoryIAM(t, cfg)
+	srv.cfg.AllowInMemory = false

 	req := httptest.NewRequest(http.MethodGet, "/api/v1/projects", nil)
 	req.Header.Set("X-User-ID", "1")
@@ -84,7 +78,7 @@ func TestP0_HealthDegradedWhenDBRequiredButUnavailable(t *testing.T) {
 		t.Fatal("expected NewService error when DB unavailable")
 	}
 	_ = svc
-	srv := NewServer(cfg, bizdev.MustNewService(t, &config.Config{DataDir: t.TempDir(), AgentBackend: "stub"}))
+	srv := newServerWithMemoryIAM(t, &config.Config{DataDir: t.TempDir(), AgentBackend: "stub"})
 	srv.cfg = cfg

 	req := httptest.NewRequest(http.MethodGet, "/api/v1/health", nil)
@@ -101,12 +95,8 @@ func TestP0_HealthDegradedWhenDBRequiredButUnavailable(t *testing.T) {
 }

 func TestP0_BodyUserIDCannotOverridePrincipal(t *testing.T) {
-	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", Env: "dev", AllowInMemory: false}
-	svc, err := bizdev.NewService(cfg)
-	if err != nil {
-		t.Fatalf("NewService: %v", err)
-	}
-	srv := NewServer(cfg, svc)
+	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", Env: "dev"}
+	srv := newServerWithMemoryIAM(t, cfg)
 	token := loginTestAdmin(t, srv)

 	body := bytes.NewReader([]byte(`{"session_id":"p0-sess","message":"hello","user_id":99999}`))
@@ -127,11 +117,7 @@ func TestP0_CORSAllowlist(t *testing.T) {
 		Env:                "dev",
 		CORSAllowedOrigins: []string{"https://app.example.com"},
 	}
-	svc, err := bizdev.NewService(cfg)
-	if err != nil {
-		t.Fatalf("NewService: %v", err)
-	}
-	srv := NewServer(cfg, svc)
+	srv := newServerWithMemoryIAM(t, cfg)

 	req := httptest.NewRequest(http.MethodOptions, "/api/v1/projects", nil)
 	req.Header.Set("Origin", "https://evil.example.com")
@@ -151,12 +137,9 @@ func TestP0_CORSAllowlist(t *testing.T) {
 }

 func TestP0_ReconcileRejectsClientWorkDir(t *testing.T) {
-	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub", AllowInMemory: false}
-	svc, err := bizdev.NewService(cfg)
-	if err != nil {
-		t.Fatalf("NewService: %v", err)
-	}
-	srv := NewServer(cfg, svc)
+	cfg := &config.Config{DataDir: t.TempDir(), AgentBackend: "stub"}
+	srv := newServerWithMemoryIAM(t, cfg)
+	srv.cfg.AllowInMemory = false

 	body := bytes.NewReader([]byte(`{"work_dir":"/tmp/evil"}`))
 	req := httptest.NewRequest(http.MethodPost, "/api/v1/admin/workspaces/reconcile", body)
@@ -14,19 +14,17 @@ import (
 )

 // newStrictAuthServer 禁用 legacy 匿名 principal，用于安全负例测试。
+// IAM 仍用 test+AllowInMemory 双门启动；启动后再关闭 AllowInMemory 以强制请求鉴权。
 func newStrictAuthServer(t *testing.T) *Server {
 	t.Helper()
 	cfg := &config.Config{
-		Env:           "main",
-		DataDir:       t.TempDir(),
-		AgentBackend:  "stub",
-		AllowInMemory: false,
+		Env:          "test",
+		DataDir:      t.TempDir(),
+		AgentBackend: "stub",
 	}
-	svc, err := bizdev.NewService(cfg)
-	if err != nil {
-		t.Fatalf("NewService: %v", err)
-	}
-	return NewServer(cfg, svc)
+	srv := newServerWithMemoryIAM(t, cfg)
+	srv.cfg.AllowInMemory = false
+	return srv
 }

 func TestPenetration_AnonymousAdminTemplateUpload401(t *testing.T) {
@@ -178,7 +176,7 @@ func TestPenetration_ReadyStrictWhenDBConfiguredButDown(t *testing.T) {
 		ContextDbPort: 59999,
 		AllowInMemory: false,
 	}
-	srv := NewServer(cfg, bizdev.MustNewService(t, &config.Config{DataDir: t.TempDir(), AgentBackend: "stub"}))
+	srv := newServerWithMemoryIAM(t, &config.Config{DataDir: t.TempDir(), AgentBackend: "stub"})
 	srv.cfg = cfg

 	req := httptest.NewRequest(http.MethodGet, "/api/v1/ready", nil)
@@ -195,7 +193,7 @@ func TestPenetration_DisallowedOriginNoWildcardCORS(t *testing.T) {
 		AgentBackend:       "stub",
 		CORSAllowedOrigins: []string{"https://trusted.example.com"},
 	}
-	srv := NewServer(cfg, bizdev.MustNewService(t, cfg))
+	srv := newServerWithMemoryIAM(t, cfg)

 	req := httptest.NewRequest(http.MethodOptions, "/api/v1/projects", nil)
 	req.Header.Set("Origin", "https://attacker.example.com")
@@ -44,6 +44,28 @@ func newTestServer(t testing.TB) *Server {
 	return srv
 }

+// newServerWithMemoryIAM 使用显式 dev/test + AllowInMemory 双门启动内存 IAM。
+// 需要「严格请求鉴权」的用例可在返回后将 srv.cfg.AllowInMemory 设为 false。
+func newServerWithMemoryIAM(t testing.TB, cfg *config.Config) *Server {
+	t.Helper()
+	if cfg == nil {
+		cfg = &config.Config{}
+	}
+	if cfg.DataDir == "" {
+		cfg.DataDir = t.TempDir()
+	}
+	if cfg.AgentBackend == "" {
+		cfg.AgentBackend = "stub"
+	}
+	cfg.AllowInMemory = true
+	env := strings.ToLower(strings.TrimSpace(cfg.Env))
+	if env == "" || env == "main" || env == "prod" {
+		cfg.Env = "test"
+	}
+	svc := bizdev.MustNewService(t, cfg)
+	return NewServer(cfg, svc)
+}
+
 func TestServer_Health(t *testing.T) {
 	srv := newTestServer(t)
 	req := httptest.NewRequest(http.MethodGet, "/api/v1/health", nil)
@@ -5,22 +5,17 @@ import (
 	"net/http/httptest"
 	"testing"

-	"chat-agent-system/internal/bizdev"
 	"chat-agent-system/internal/config"
 )

 func TestServer_WebAssetsHostingAndFallback(t *testing.T) {
 	cfg := &config.Config{
-		Env:           "test",
-		AgentBackend:  "stub",
-		DataDir:       t.TempDir(),
-		AllowInMemory: false,
+		Env:          "test",
+		AgentBackend: "stub",
+		DataDir:      t.TempDir(),
 	}
-	svc, err := bizdev.NewService(cfg)
-	if err != nil {
-		t.Fatalf("NewService: %v", err)
-	}
-	srv := NewServer(cfg, svc)
+	srv := newServerWithMemoryIAM(t, cfg)
+	srv.cfg.AllowInMemory = false
 	handler := srv.Routes()

 	reqRoot := httptest.NewRequest(http.MethodGet, "/", nil)