Merge pull request 'Update Gitea endpoints after server migration' (#56) from fix/gitea-endpoint-migration into main

Why: recursive submodule checkout reads pinned dep commits; their
.gitmodules now carries the new Gitea endpoint base only in the
migrated mains.

Evidence: host-verified; see Dev PR #706

Impact: pin updates only.

Refs: Agent-System-Dev PR #706

.gitea/workflows/ci.yml

@@ -24,7 +24,7 @@ jobs:
             echo "SUBMODULE_TOKEN is not configured" >&2
             exit 1
           fi
-          git config --global url."https://weichen:${SUBMODULE_TOKEN}@git-home.u1680832.nyat.app:30319/".insteadOf "ssh://git@git-ssh.u1680832.nyat.app:31411/"
+          git config --global url."https://weichen:${SUBMODULE_TOKEN}@gitea.naitangbaobao.cn/".insteadOf "ssh://git@gitea.naitangbaobao.cn:52222/"
           git -c fetch.recurseSubmodules=false submodule update --init --recursive
       - name: Install Go 1.27.1
         run: |

.gitea/workflows/governance.yml

@@ -31,7 +31,7 @@ jobs:
           parsed_server = urlsplit(SERVER)
           if parsed_server.scheme not in {"http", "https"} or not parsed_server.netloc:
               sys.exit("Checkout server must be the platform-injected http(s) URL")
-          SSH = "ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/"
+          SSH = "ssh://git@gitea.naitangbaobao.cn:52222/weichen/"
           # 审查过的白名单，独立于候选 .gitmodules：仅列出两个治理 gitlink。
           REPOSITORIES = {
               "deps/development-guidelines": "Development-Guidelines",

.gitmodules

@@ -1,18 +1,18 @@
 [submodule "deps/agent-native"]
 	path = deps/agent-native
-	url = ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/Agent-Native.git
+	url = ssh://git@gitea.naitangbaobao.cn:52222/weichen/Agent-Native.git
 [submodule "deps/development-guidelines"]
 	path = deps/development-guidelines
-	url = ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/Development-Guidelines.git
+	url = ssh://git@gitea.naitangbaobao.cn:52222/weichen/Development-Guidelines.git
 [submodule "deps/error-code-system"]
 	path = deps/error-code-system
-	url = ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/Error-Code-System.git
+	url = ssh://git@gitea.naitangbaobao.cn:52222/weichen/Error-Code-System.git
 [submodule "deps/logging-system"]
 	path = deps/logging-system
-	url = ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/Logging-System.git
+	url = ssh://git@gitea.naitangbaobao.cn:52222/weichen/Logging-System.git
 [submodule "deps/agent-intent"]
 	path = deps/agent-intent
-	url = ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/Agent-Intent.git
+	url = ssh://git@gitea.naitangbaobao.cn:52222/weichen/Agent-Intent.git
 [submodule "deps/development-manager"]
 	path = deps/development-manager
-	url = ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/Development-Manager.git
+	url = ssh://git@gitea.naitangbaobao.cn:52222/weichen/Development-Manager.git

AGENT-EXECUTION.md

@@ -36,19 +36,13 @@ profiles:
     adapter: cursor
     executable: "cursor-agent-stable"
     model: auto
-    # 2026-10-21 临时禁用：U3B 连续两次 fail-stop（首次 TLS 会话前故障 553B 日志、
-    # 二次 "Cannot use this model: auto"，与 Backend 配置已记录的月配额耗尽一致），
-    # 两次均未建立会话、零改动（现场保留）。恢复时改回 enabled: true。
-    enabled: false
+    enabled: true
 
   antigravity:
     adapter: antigravity
     executable: agy
     model: auto
-    # 2026-10-21 临时禁用：U3B 尝试被 400 FAILED_PRECONDITION "User location is
-    # not supported for the API use" 拒绝，0 token、未建立会话（现场保留）；
-    # 与 Backend 配置已记录的 Gemini 端点不稳一致。恢复时改回 enabled: true。
-    enabled: false
+    enabled: true
 
   pi:
     adapter: pi

AGENTS.md

@@ -23,9 +23,9 @@
 历史文档与旧链接里的 `git-panel.u1680832.nyat.app:30319` 已失效（TLS 握手失败），
 见到一律按过期处理，不要再试。当前权威通道：
 
-- **Web / API（HTTPS）**：`https://git-home.u1680832.nyat.app:30319`（Gitea 1.27.3，
+- **Web / API（HTTPS）**：`https://gitea.naitangbaobao.cn`（Gitea 1.27.3，
   API 前缀 `/api/v1`；偶发 broken pipe / 超时，重试即可）。
-- **Git over SSH**：`ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/<Repo>.git`
+- **Git over SSH**：`ssh://git@gitea.naitangbaobao.cn:52222/weichen/<Repo>.git`
   （各仓 origin 与 gitlink 均为此格式）。
 - **API token**：`~/.secrets/gitea`，两行键值 `URL_PREFIX=…` 与
   `ACCESS_TOKEN=<40 位 hex>`；取值 `grep ACCESS_TOKEN ~/.secrets/gitea | cut -d= -f2-`，
@@ -33,7 +33,7 @@
   （baseline 硬边界 2：只记录来源、字段与可用状态）。
 - **SSH key**：`~/.ssh/home_gitea`（ED25519，公钥指纹
   `SHA256:GYXaOXIMeloQ1fKSfdJ/ndgPQzs/Ix7Hk62ekfkfQ20`，Gitea 侧密钥名 GMKtec）；
-  `~/.ssh/config` 已有 `git-ssh.u1680832.nyat.app` 主机条目，克隆/推送直接可用。
+  `~/.ssh/config` 已有 `gitea.naitangbaobao.cn（端口 52222）` 主机条目，克隆/推送直接可用。
 
 ## Agent 执行与新项目准入（强制）
 

PROJECT.md

@@ -6,7 +6,7 @@
 
 - 目标/产品边界：通用智能体小队编排底座。角色、蓝图、规划、执行引擎、黑板、HITL 与 HTTP/CLI 组合根。不含预设医疗器械角色或任务模板。不实现 Agent。
 - 源码、资产、文档、机器契约位置：Go 模块路径 `agent-squad`。包为 `squad`、`store`、`server`、`cli`、`textextract`、`conformance`。人类文档在根 `README.md`、`policy/`；过程记录在 `docs/records/`。对外 HTTP 错误合同以错误码库信封为准。
-- 权威远端、集成基线、受保护分支：`ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/Agent-Squad.git`，集成基线 `main`。
+- 权威远端、集成基线、受保护分支：`ssh://git@gitea.naitangbaobao.cn:52222/weichen/Agent-Squad.git`，集成基线 `main`。
 - 并行共同基线记录位置、合入通知/主动检查方式：Gitea Issue/PR。跨仓共同起点是本仓提交加上它锁定的 Agent-Native、错误码库、日志库与规范库 SHA。
 - 发布线与批准版本/实际制品查询入口：无独立制品库。可执行文件由 `go build` 从固定 gitlink 组合产出。
 - 提交/审核/合并/发布身份与权限：仓库所有者 weichen。
@@ -33,7 +33,7 @@
 克隆：
 
 ```bash
-git clone --recurse-submodules ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/Agent-Squad.git
+git clone --recurse-submodules ssh://git@gitea.naitangbaobao.cn:52222/weichen/Agent-Squad.git
 # 已克隆时：git submodule update --init --recursive
 ```
 

README.md

@@ -21,7 +21,7 @@ Agent-Native 自己的 `deps/` 里还有一份错误码库、日志库和规范
 克隆时带上子模块，否则无法编译。已有克隆若仍留着根目录 `Agent-Native`，先删掉该残留目录再初始化：
 
 ```bash
-git clone --recurse-submodules ssh://git@git-ssh.u1680832.nyat.app:31411/weichen/Agent-Squad.git
+git clone --recurse-submodules ssh://git@gitea.naitangbaobao.cn:52222/weichen/Agent-Squad.git
 # 已克隆时：
 git submodule sync --recursive
 git submodule update --init --recursive

adapters/native/register.go

@@ -3,13 +3,8 @@ package native
 
 import (
 	"context"
-	"crypto/sha256"
-	"encoding/hex"
-	"encoding/json"
 	"fmt"
-	"reflect"
 	"strings"
-	"sync"
 	"time"
 
 	nativexec "agent-native/adapters/native"
@@ -57,58 +52,12 @@ type Config struct {
 	PermissionAsk func(ctx context.Context, call llm.ToolCall) runtime.AskVerdict
 }
 
-// Identity 返回塑造 Executor 的配置摘要（模型/客户端/工具/Exec 等）。
-// RegisterFull 经 BackendResolver.Register 替换工厂时会失效缓存；角色侧
-// BashPermission 等由 squad.ConfigIdentity 纳入 Resolve 缓存键，故 allow→deny
-// 在下一次 Resolve 即生效，无需重启进程。
-func (c Config) Identity() string {
-	var b strings.Builder
-	b.WriteString("model=")
-	b.WriteString(c.DefaultModel)
-	b.WriteString("\x00turns=")
-	b.WriteString(fmt.Sprintf("%d", c.MaxTurns))
-	b.WriteString("\x00timeout=")
-	b.WriteString(c.Timeout.String())
-	if c.Client != nil {
-		b.WriteString("\x00base=")
-		b.WriteString(c.Client.BaseURL)
-		b.WriteString("\x00key=")
-		b.WriteString(c.Client.APIKey)
-	}
-	if c.Exec != nil {
-		b.WriteString("\x00exec.bash=")
-		b.WriteString(c.Exec.Bash)
-		b.WriteString("\x00exec.python=")
-		b.WriteString(c.Exec.Python)
-	}
-	if c.Search != nil {
-		b.WriteString("\x00search=1")
-	}
-	if c.Fetch != nil {
-		b.WriteString("\x00fetch=1")
-	}
-	if c.MCP != nil {
-		b.WriteString("\x00mcp=1")
-	}
-	if len(c.SkillsRoots) > 0 {
-		b.WriteString("\x00skills=")
-		b.WriteString(strings.Join(c.SkillsRoots, ","))
-	}
-	if c.CodeOrchestration {
-		b.WriteString("\x00code=1")
-	}
-	sum := sha256.Sum256([]byte(b.String()))
-	return hex.EncodeToString(sum[:16])
-}
-
 // Register 把 "agent-native" 工厂注册进解析器。
 func Register(r *squad.BackendResolver, client *llm.Client, defaultModel string, maxTurns int, timeout time.Duration) {
 	RegisterFull(r, Config{Client: client, DefaultModel: defaultModel, MaxTurns: maxTurns, Timeout: timeout})
 }
 
 // RegisterFull 注册带工具装配的 "agent-native" 工厂。
-// 再次调用会替换工厂并失效该后端缓存（见 BackendResolver.Register）；
-// 配置摘要见 Config.Identity，角色侧见 squad.ConfigIdentity。
 func RegisterFull(r *squad.BackendResolver, cfg Config) {
 	r.Register("agent-native", func(role squad.RoleSpec) (squad.Executor, error) {
 		opts, err := buildOptions(cfg, role)
@@ -185,19 +134,12 @@ func toNative(req squad.ExecRequest) exec.Request {
 	}
 	if req.OnEvent != nil {
 		prev := req.OnEvent
-		// 一帧一回调：同步转发，不缓存、不批量重发、不按调用分组——下游看到的
-		// 就是 native 的到达序。富化在原帧上就地完成（逐帧独立取值），
-		// 并行工具调用互不串字段，也就不会把交错帧重排成按调用分组。
 		out.OnEvent = func(ev exec.ProcessEvent) {
-			// CallID/Status/ErrorReason 是 native 侧的可选字段（U11a 落地）：
-			// 缺席 = legacy，取零值透传，绝不伪造。
-			extras := probeNativeEventExtras(ev)
 			prev(squad.ProcessEvent{
 				Type: ev.Type, Text: ev.Text, Tool: ev.Tool, Detail: ev.Detail,
 				PromptTokens: ev.PromptTokens, CompletionTokens: ev.CompletionTokens,
 				LastPromptTokens: ev.LastPromptTokens, LastCompletionTokens: ev.LastCompletionTokens,
 				CacheReadTokens: ev.CacheReadTokens, CacheWriteTokens: ev.CacheWriteTokens,
-				CallID: extras.CallID, Status: extras.Status, ErrorReason: extras.ErrorReason,
 			})
 		}
 	}
@@ -205,148 +147,28 @@ func toNative(req squad.ExecRequest) exec.Request {
 }
 
 func fromNative(res *exec.Result) *squad.ExecResult {
-	if res == nil {
-		return nil
-	}
-	out := &squad.ExecResult{
-		Success:          res.Success,
-		Patches:          res.Patches,
-		Logs:             res.Logs,
-		ErrorReason:      res.ErrorReason,
-		Retryable:        res.Retryable,
-		PromptTokens:     res.PromptTokens,
-		CompletionTokens: res.CompletionTokens,
-	}
-	// F05/U3b（Squad 侧 contract）：
-	//   - native exec.Result 在后续 release 会新增 SideEffects、Canceled、Aborted 字段；
-	//     本地 deps gitlink 若尚未跟进，从 JSON 探测拿不到这些字段时回落到
-	//     「legacy reading: side effects = possible」，与既有调用方旧行为一致。
-	//   - 在不影响 ExecResult 默认值的前提下，用 JSON 探测而非硬性修改 deps。
-	probe, err := probeNativeResultExtras(res)
-	if err == nil {
-		out.SideEffects = probe.SideEffects
-		out.Canceled = probe.Canceled
-		out.Aborted = probe.Aborted
-	}
-	return out
-}
-
-// nativeResultExtras 探测 native exec.Result 上的可选字段（结构化 JSON）。
-// 缺省（字段不存在或为零值时）按 legacy reading：SideEffects 空串视为 "possible"；
-// 调用方侧 ExecResult.IsRetryableSideEffectReports() 已经处理该回退。
-type nativeResultExtras struct {
-	SideEffects squad.ExecSideEffects `json:"side_effects,omitempty"`
-	Canceled    bool                  `json:"canceled,omitempty"`
-	Aborted     bool                  `json:"aborted,omitempty"`
-}
-
-// probeNativeResultExtras 通过 JSON 探测 native Result 上的可选字段。
-// 探测出 cancel/abort/side_effects 任何一个就使用它们；探测失败或全部为零值视为
-// legacy reading（SideEffects 空 = possible 兜底）。
-func probeNativeResultExtras(res *exec.Result) (nativeResultExtras, error) {
-	if res == nil {
-		return nativeResultExtras{}, fmt.Errorf("nil result")
-	}
-	v := reflect.ValueOf(res).Elem()
-	if !v.IsValid() {
-		return nativeResultExtras{}, fmt.Errorf("invalid result")
-	}
-	// 用 JSON 做结构化探测：这一路兼容 deps 升级（增字段）、不增字段两种情形。
-	data, err := json.Marshal(v.Interface())
-	if err != nil {
-		return nativeResultExtras{}, err
-	}
-	var out nativeResultExtras
-	if err := json.Unmarshal(data, &out); err != nil {
-		return nativeResultExtras{}, err
-	}
-	return out, nil
-}
-
-// nativeEventExtras 是 native exec.ProcessEvent 上的可选字段：工具调用身份与
-// 真实结局（U11b，语义与 native 侧一致：空串 = legacy/未知）。
-type nativeEventExtras struct {
-	CallID      string
-	Status      string
-	ErrorReason string
-}
-
-// probeNativeEventExtras 读取 native 过程帧上的 CallID/Status/ErrorReason。
-//
-// 与 probeNativeResultExtras 同一取法（可选字段缺席 = legacy 零值），但改用
-// 反射按字段名读取而不是 JSON：本地 deps gitlink 上的 exec.ProcessEvent 尚未
-// 跟进这三个字段（U11a 落地），硬写 ev.CallID 会编译不过；反射让同一份代码
-// 两侧都成立——字段缺席取零值（既有行为逐字节不变），字段落地后无需改本文件
-// 即生效，也不受 native 是否给字段加 JSON tag 影响。每次过程帧只做三次字段
-// 定位，不构造中间对象；native 可并行发帧时读到的仍是该帧自己的值。
-func probeNativeEventExtras(v any) nativeEventExtras {
-	rv := reflect.ValueOf(v)
-	for rv.Kind() == reflect.Pointer || rv.Kind() == reflect.Interface {
-		if rv.IsNil() {
-			return nativeEventExtras{}
-		}
-		rv = rv.Elem()
-	}
-	if rv.Kind() != reflect.Struct {
-		return nativeEventExtras{}
-	}
-	return nativeEventExtras{
-		CallID:      optionalStringField(rv, "CallID"),
-		Status:      optionalStringField(rv, "Status"),
-		ErrorReason: optionalStringField(rv, "ErrorReason"),
+	return &squad.ExecResult{
+		Success: res.Success, Patches: res.Patches, Logs: res.Logs, ErrorReason: res.ErrorReason,
+		Retryable: res.Retryable, PromptTokens: res.PromptTokens, CompletionTokens: res.CompletionTokens,
 	}
 }
 
-// optionalStringField 按字段名取字符串字段；缺席、嵌套或类型不符都返回 ""
-// （legacy 零值）。只取直接字段，不做嵌套指针下钻。
-func optionalStringField(rv reflect.Value, name string) string {
-	field, ok := rv.Type().FieldByName(name)
-	if !ok || len(field.Index) != 1 {
-		return ""
-	}
-	value := rv.Field(field.Index[0])
-	if value.Kind() != reflect.String {
-		return ""
-	}
-	return value.String()
-}
-
-// eventTap 在组合根侧再挂一层过程事件观察。帧按值传递（squad.ProcessEvent
-// 只含值类型字段），两个下游各自拿到整帧副本，并行工具调用不会互相串字段；
-// 回调保持同步，不排队、不缓存。
-//
-// 扇出以 mu 串行：「一帧 → 全部下游」是一个整体。这不是可有可无的细节：
-// native 对并行工具调用是每调用一个 goroutine 发帧，两次帧到达可以真正并发。
-// 不串行时，帧 F 交给组合根侧 sink 之后、在交给请求侧 sink 之前，可能被后到
-// 的帧 G 抢先递完，于是同一串到达帧在两个下游呈现出不同的顺序——实测就是
-// 组合根侧看到 [start(err), start(ok), end(ok), end(err)]（到达序），而请求侧
-// 看到 [start(ok), end(ok), start(err), end(err)]（被按调用重排/分组）。
-// 串行扇出后每个下游看到的都是同一串到达序：一个上游帧 → 每个下游恰好一次、
-// 顺序不变，且事件富化（CallID/Status/ErrorReason）在桥里就地完成、不在这里
-// 重新分帧。mu 只在本次扇出期间持有，不跨帧持有。
 type eventTap struct {
 	next squad.Executor
 	on   func(squad.ProcessEvent)
-	mu   sync.Mutex
 }
 
 func (b *eventTap) Name() string { return b.next.Name() }
 
 func (b *eventTap) Execute(ctx context.Context, req squad.ExecRequest) (*squad.ExecResult, error) {
-	req.OnEvent = b.hook(req.OnEvent)
-	return b.next.Execute(ctx, req)
-}
-
-// hook 把一帧扇出到组合根侧 sink 与请求侧 sink，顺序不变。
-func (b *eventTap) hook(prev func(squad.ProcessEvent)) func(squad.ProcessEvent) {
-	return func(ev squad.ProcessEvent) {
-		b.mu.Lock()
-		defer b.mu.Unlock()
+	prev := req.OnEvent
+	req.OnEvent = func(ev squad.ProcessEvent) {
 		b.on(ev)
 		if prev != nil {
 			prev(ev)
 		}
 	}
+	return b.next.Execute(ctx, req)
 }
 
 // LLM 把 llm.Client 适配成 ModelClient。

adapters/native/register_event_test.go

@@ -1,427 +0,0 @@
-package native
-
-import (
-	"context"
-	"reflect"
-	"sync"
-	"testing"
-	"time"
-
-	"agent-native/exec"
-	"agent-squad/squad"
-)
-
-// nativeEventWithIDs 模拟 U11a 落地后的 native exec.ProcessEvent 形态：
-// 多出的 CallID / Status / ErrorReason，且带 snake_case JSON tag
-// （证明 squad 侧读取不受 native 的 tag 写法影响）。
-type nativeEventWithIDs struct {
-	Type        string `json:"type"`
-	Text        string `json:"text"`
-	Tool        string `json:"tool"`
-	Detail      string `json:"detail"`
-	CallID      string `json:"call_id,omitempty"`
-	Status      string `json:"status,omitempty"`
-	ErrorReason string `json:"error_reason,omitempty"`
-}
-
-func TestBridgeMapsLegacyNativeEventWithoutFabricatingFields(t *testing.T) {
-	var got []squad.ProcessEvent
-	out := toNative(squad.ExecRequest{OnEvent: func(ev squad.ProcessEvent) { got = append(got, ev) }})
-	if out.OnEvent == nil {
-		t.Fatal("bridge must install its own event hook when the request has one")
-	}
-	out.OnEvent(exec.ProcessEvent{
-		Type: "tool_execution_end", Text: "found", Tool: "grep", Detail: `{"q":"x"}`,
-		PromptTokens: 11, CompletionTokens: 22,
-		LastPromptTokens: 3, LastCompletionTokens: 4, CacheReadTokens: 5, CacheWriteTokens: 6,
-	})
-	if len(got) != 1 {
-		t.Fatalf("want 1 squad event, got %d", len(got))
-	}
-	want := squad.ProcessEvent{
-		Type: "tool_execution_end", Text: "found", Tool: "grep", Detail: `{"q":"x"}`,
-		PromptTokens: 11, CompletionTokens: 22,
-		LastPromptTokens: 3, LastCompletionTokens: 4, CacheReadTokens: 5, CacheWriteTokens: 6,
-	}
-	if !reflect.DeepEqual(got[0], want) {
-		t.Fatalf("legacy native event must map byte-identically:\n got %+v\nwant %+v", got[0], want)
-	}
-	// 无请求侧回调时不装钩子（既有行为不变）。
-	if toNative(squad.ExecRequest{}).OnEvent != nil {
-		t.Fatal("bridge must not install an event hook when the request has none")
-	}
-}
-
-func TestProbeNativeEventExtrasReadsNewFieldsWhenNativeHasThem(t *testing.T) {
-	// legacy 帧：字段缺席（老 pin）或在场但未填（当前 pin 的零值）→ 一律零值，
-	// 不臆造身份与结局。
-	if extras := probeNativeEventExtras(exec.ProcessEvent{Type: "tool_execution_end", Tool: "grep"}); extras != (nativeEventExtras{}) {
-		t.Fatalf("legacy native event must yield zero extras, got %+v", extras)
-	}
-	// U11a 落地后：按字段名读到真实值（与 JSON tag 写法无关）。
-	future := nativeEventWithIDs{Type: "tool_execution_end", Tool: "grep", CallID: "call_err",
-		Status: "error", ErrorReason: "grep: permission denied"}
-	want := nativeEventExtras{CallID: "call_err", Status: "error", ErrorReason: "grep: permission denied"}
-	if extras := probeNativeEventExtras(future); extras != want {
-		t.Fatalf("identity/outcome not carried through: %+v", extras)
-	}
-	if extras := probeNativeEventExtras(&future); extras != want {
-		t.Fatalf("pointer and value frames must probe identically: %+v", extras)
-	}
-	// 缺席/非字符串字段都回落到零值。
-	odd := struct {
-		CallID      int
-		Status      string
-		ErrorReason bool
-	}{CallID: 7, Status: "ok", ErrorReason: true}
-	if extras := probeNativeEventExtras(odd); extras != (nativeEventExtras{Status: "ok"}) {
-		t.Fatalf("non-string/absent fields must stay zero-valued, got %+v", extras)
-	}
-	var nilFrame *nativeEventWithIDs
-	if extras := probeNativeEventExtras(nilFrame); extras != (nativeEventExtras{}) {
-		t.Fatalf("nil frame must stay zero-valued, got %+v", extras)
-	}
-	if extras := probeNativeEventExtras("not a frame"); extras != (nativeEventExtras{}) {
-		t.Fatalf("non-struct input must stay zero-valued, got %+v", extras)
-	}
-	// 嵌套（嵌入）字段不探测：不沿 nil 嵌入指针下钻，也不当作直接字段读取。
-	nested := struct {
-		*nativeEventWithIDs
-		Status string
-	}{Status: "ok"}
-	if extras := probeNativeEventExtras(nested); extras != (nativeEventExtras{Status: "ok"}) {
-		t.Fatalf("nested optional fields must stay zero-valued, got %+v", extras)
-	}
-}
-
-// interleavedToolFrames 是并行工具调用的固定交错帧序（squad 侧形态）：
-//
-//	start(call_err) → start(call_ok) → end(call_ok, ok) → end(call_err, error)
-//
-// 按到达序转发时逐帧等值；任何「按调用分组重排」「按调用缓存后再发」都会把它
-// 变成 [start, end, start, end]，因此它能把重排/重发类缺陷钉成硬失败。
-func interleavedToolFrames() []squad.ProcessEvent {
-	return []squad.ProcessEvent{
-		{Type: "tool_execution_start", Tool: "grep", CallID: "call_err"},
-		{Type: "tool_execution_start", Tool: "grep", CallID: "call_ok"},
-		{Type: "tool_execution_end", Tool: "grep", CallID: "call_ok",
-			Status: squad.ToolStatusOK, Text: "found"},
-		{Type: "tool_execution_end", Tool: "grep", CallID: "call_err",
-			Status: squad.ToolStatusError, ErrorReason: "grep: permission denied"},
-	}
-}
-
-// interleavedNativeFrames 是 interleavedToolFrames 的 native 侧形态：U11a 落地后
-// （当前 pin 即此形态）的帧自带 CallID/Status/ErrorReason，桥读到的就是它们。
-func interleavedNativeFrames() []exec.ProcessEvent {
-	return []exec.ProcessEvent{
-		{Type: "tool_execution_start", Tool: "grep", CallID: "call_err"},
-		{Type: "tool_execution_start", Tool: "grep", CallID: "call_ok"},
-		{Type: "tool_execution_end", Tool: "grep", CallID: "call_ok",
-			Status: squad.ToolStatusOK, Text: "found"},
-		{Type: "tool_execution_end", Tool: "grep", CallID: "call_err",
-			Status: squad.ToolStatusError, ErrorReason: "grep: permission denied"},
-	}
-}
-
-// legacyToolFrames 是同一个交错序列的 legacy squad 形态：身份与结局字段为空串
-// （上游未填充），失败真相只在 Text 里——空串 = 未知，不是「无名调用」。
-func legacyToolFrames() []squad.ProcessEvent {
-	return []squad.ProcessEvent{
-		{Type: "tool_execution_start", Tool: "grep"},
-		{Type: "tool_execution_start", Tool: "grep"},
-		{Type: "tool_execution_end", Tool: "grep", Text: "found"},
-		{Type: "tool_execution_end", Tool: "grep", Text: "grep: permission denied"},
-	}
-}
-
-// legacyNativeFrames 是 legacy native 事件流：帧上没有 CallID/Status/ErrorReason
-// （字段不存在，或存在但未填——对探测而言都是零值）。桥必须原样透传。
-func legacyNativeFrames() []exec.ProcessEvent {
-	return []exec.ProcessEvent{
-		{Type: "tool_execution_start", Tool: "grep"},
-		{Type: "tool_execution_start", Tool: "grep"},
-		{Type: "tool_execution_end", Tool: "grep", Text: "found"},
-		{Type: "tool_execution_end", Tool: "grep", Text: "grep: permission denied"},
-	}
-}
-
-// assertArrivalOrder 断言帧序逐帧等于到达序：一个上游帧对应下游一帧，顺序、
-// 身份与结局全部逐帧等值（squad.ProcessEvent 只含值类型字段，可直接比较）。
-func assertArrivalOrder(t *testing.T, tag string, got, want []squad.ProcessEvent) {
-	t.Helper()
-	if len(got) != len(want) {
-		t.Fatalf("%s: frames = %d, want %d (每帧一次转发，不多不少不重发):\n%+v", tag, len(got), len(want), got)
-	}
-	for i := range want {
-		if got[i] != want[i] {
-			t.Fatalf("%s: frame %d = %+v\nwant %+v\n（帧序必须与到达序逐帧一致）", tag, i, got[i], want[i])
-		}
-	}
-}
-
-// concurrentToolExecutor 并发发两次同名工具调用的过程帧（CallID 不同、一成一败），
-// 模拟 native 侧并行工具执行时的帧序交错。
-type concurrentToolExecutor struct{}
-
-func (concurrentToolExecutor) Name() string { return "concurrent-tool" }
-
-func (concurrentToolExecutor) Execute(ctx context.Context, req squad.ExecRequest) (*squad.ExecResult, error) {
-	calls := []struct {
-		callID, status, errorReason, text string
-	}{
-		{callID: "call_ok", status: squad.ToolStatusOK, text: "found"},
-		{callID: "call_err", status: squad.ToolStatusError, errorReason: "grep: permission denied"},
-	}
-	var wg sync.WaitGroup
-	for _, call := range calls {
-		wg.Add(1)
-		go func(callID, status, errorReason, text string) {
-			defer wg.Done()
-			if req.OnEvent == nil {
-				return
-			}
-			req.OnEvent(squad.ProcessEvent{Type: "tool_execution_start", Tool: "grep", CallID: callID})
-			req.OnEvent(squad.ProcessEvent{Type: "tool_execution_end", Tool: "grep", CallID: callID,
-				Status: status, ErrorReason: errorReason, Text: text})
-		}(call.callID, call.status, call.errorReason, call.text)
-	}
-	wg.Wait()
-	return &squad.ExecResult{Success: true}, nil
-}
-
-// assertPaired 断言按 CallID 配对：每个 call_id 一帧 start、一帧 end，
-// 结局属于该调用自身（并行工具调用不串字段）。
-func assertPaired(t *testing.T, tag string, frames []squad.ProcessEvent) {
-	t.Helper()
-	starts := map[string]int{}
-	ends := map[string]squad.ProcessEvent{}
-	for _, ev := range frames {
-		if ev.CallID == "" {
-			t.Fatalf("%s: frame without CallID: %+v", tag, ev)
-		}
-		switch ev.Type {
-		case "tool_execution_start":
-			starts[ev.CallID]++
-		case "tool_execution_end":
-			if _, dup := ends[ev.CallID]; dup {
-				t.Fatalf("%s: duplicate end for %s: %+v", tag, ev.CallID, ev)
-			}
-			ends[ev.CallID] = ev
-		}
-	}
-	if len(ends) != 2 {
-		t.Fatalf("%s: distinct CallIDs = %d, want 2 (%+v)", tag, len(ends), frames)
-	}
-	for _, want := range []struct {
-		callID, status, errorReason string
-	}{
-		{callID: "call_ok", status: squad.ToolStatusOK},
-		{callID: "call_err", status: squad.ToolStatusError, errorReason: "grep: permission denied"},
-	} {
-		if starts[want.callID] != 1 {
-			t.Fatalf("%s: call %s starts = %d, want 1", tag, want.callID, starts[want.callID])
-		}
-		end := ends[want.callID]
-		if end.Status != want.status || end.ErrorReason != want.errorReason {
-			t.Fatalf("%s: call %s outcome = %s/%q, want %s/%q", tag, want.callID,
-				end.Status, end.ErrorReason, want.status, want.errorReason)
-		}
-	}
-}
-
-// TestEventTapKeepsConcurrentToolCallIdentityAndOutcome 在真实并发帧源下验证三件事：
-// 身份与结局不串（严格按 CallID 配对）、帧不多发不少发（每个调用一 start 一 end）、
-// 所有下游观察到同一串到达序（并发帧源下到达序本身不确定，但必须唯一且一致）。
-func TestEventTapKeepsConcurrentToolCallIdentityAndOutcome(t *testing.T) {
-	var tapped, requested []squad.ProcessEvent
-	var mu sync.Mutex
-	tap := &eventTap{next: concurrentToolExecutor{}, on: func(ev squad.ProcessEvent) {
-		mu.Lock()
-		defer mu.Unlock()
-		tapped = append(tapped, ev)
-	}}
-	_, err := tap.Execute(context.Background(), squad.ExecRequest{
-		OnEvent: func(ev squad.ProcessEvent) {
-			mu.Lock()
-			defer mu.Unlock()
-			requested = append(requested, ev)
-		},
-	})
-	if err != nil {
-		t.Fatal(err)
-	}
-	assertPaired(t, "tap", tapped)
-	assertPaired(t, "request", requested)
-	if !reflect.DeepEqual(tapped, requested) {
-		t.Fatalf("every sink must observe the same arrival order (one frame per upstream frame, unbuffered):\ntap     %+v\nrequest %+v", tapped, requested)
-	}
-}
-
-// TestNativeFramesReachBothTapSinksInArrivalOrder 走完整一跳：native 帧 → 桥的
-// OnEvent（就地富化 CallID/Status/ErrorReason）→ 事件 tap → 组合根侧 sink 与
-// 请求侧 sink。两种 native（当前 pin 的带字段帧 / legacy 无字段帧）都要逐帧
-// 等值、顺序不变：既不能被按调用分组重排，也不能缓存后再批量发。
-func TestNativeFramesReachBothTapSinksInArrivalOrder(t *testing.T) {
-	for _, tc := range []struct {
-		name   string
-		native []exec.ProcessEvent
-		want   []squad.ProcessEvent
-	}{
-		{name: "current-native", native: interleavedNativeFrames(), want: interleavedToolFrames()},
-		{name: "legacy-native", native: legacyNativeFrames(), want: legacyToolFrames()},
-	} {
-		t.Run(tc.name, func(t *testing.T) {
-			var mu sync.Mutex
-			var tapped, requested []squad.ProcessEvent
-			tap := &eventTap{on: func(ev squad.ProcessEvent) {
-				mu.Lock()
-				defer mu.Unlock()
-				tapped = append(tapped, ev)
-			}}
-			nativeHook := toNative(squad.ExecRequest{OnEvent: tap.hook(func(ev squad.ProcessEvent) {
-				mu.Lock()
-				defer mu.Unlock()
-				requested = append(requested, ev)
-			})}).OnEvent
-			if nativeHook == nil {
-				t.Fatal("bridge must install its own event hook when the request has one")
-			}
-			for _, ev := range tc.native {
-				nativeHook(ev)
-			}
-			assertArrivalOrder(t, tc.name+"/tap", tapped, tc.want)
-			assertArrivalOrder(t, tc.name+"/request", requested, tc.want)
-			if tc.name == "current-native" {
-				assertPaired(t, tc.name, tapped)
-				assertPaired(t, tc.name, requested)
-			}
-		})
-	}
-}
-
-// TestBridgeForwardsNativeFramesInArrivalOrder 直接钉住桥这一跳的帧序与富化：
-// 当前 pin（帧带 CallID/Status/ErrorReason）就地把值带上来，legacy 帧（无这三个
-// key）就以空串原样透传——两种情况都必须逐帧等值、顺序不变。
-func TestBridgeForwardsNativeFramesInArrivalOrder(t *testing.T) {
-	for _, tc := range []struct {
-		name   string
-		native []exec.ProcessEvent
-		want   []squad.ProcessEvent
-	}{
-		{name: "current-native", native: interleavedNativeFrames(), want: interleavedToolFrames()},
-		{name: "legacy-native", native: legacyNativeFrames(), want: legacyToolFrames()},
-	} {
-		t.Run(tc.name, func(t *testing.T) {
-			var got []squad.ProcessEvent
-			hook := toNative(squad.ExecRequest{OnEvent: func(ev squad.ProcessEvent) {
-				got = append(got, ev)
-			}}).OnEvent
-			if hook == nil {
-				t.Fatal("bridge must install its own event hook when the request has one")
-			}
-			for _, ev := range tc.native {
-				hook(ev)
-			}
-			assertArrivalOrder(t, tc.name, got, tc.want)
-			if tc.name == "current-native" {
-				assertPaired(t, tc.name, got)
-			}
-		})
-	}
-}
-
-// fanoutWindow 制造扇出竞态窗口：支线 A 逐帧发 call_ok 的两帧；当 A 的第二帧进入
-// 组合根侧 sink 时（entered 关闭）支线 B 才发 call_err 的两帧，并在自己的扇出
-// 全部返回后关闭 finished。A 在 sink 里等 finished，于是「A 已收下第二帧、还没
-// 交给请求侧 sink」这个窗口被固定复现——扇出不串行时 B 会抢先穿过这个窗口。
-type fanoutWindow struct {
-	entered  chan struct{}
-	finished chan struct{}
-}
-
-func (fanoutWindow) Name() string { return "fanout-window" }
-
-func (w fanoutWindow) Execute(ctx context.Context, req squad.ExecRequest) (*squad.ExecResult, error) {
-	if req.OnEvent == nil {
-		return &squad.ExecResult{Success: true}, nil
-	}
-	var wg sync.WaitGroup
-	wg.Add(2)
-	go func() {
-		defer wg.Done()
-		req.OnEvent(squad.ProcessEvent{Type: "tool_execution_start", Tool: "grep", CallID: "call_ok"})
-		req.OnEvent(squad.ProcessEvent{Type: "tool_execution_end", Tool: "grep", CallID: "call_ok",
-			Status: squad.ToolStatusOK, Text: "found"})
-	}()
-	go func() {
-		defer wg.Done()
-		<-w.entered
-		req.OnEvent(squad.ProcessEvent{Type: "tool_execution_start", Tool: "grep", CallID: "call_err"})
-		req.OnEvent(squad.ProcessEvent{Type: "tool_execution_end", Tool: "grep", CallID: "call_err",
-			Status: squad.ToolStatusError, ErrorReason: "grep: permission denied"})
-		close(w.finished)
-	}()
-	wg.Wait()
-	return &squad.ExecResult{Success: true}, nil
-}
-
-// TestEventTapFansOutOneFramePerSinkInOneOrder 把「扇出是一个整体」钉成确定性失败：
-// 组合根侧 sink 在第二帧上主动等另一条支线走完自己的扇出。没有扇出串行化时，
-// 请求侧 sink 会先收到 call_err 的两帧，组合根侧 sink 却先收到 call_ok 的两帧——
-// 同一串到达帧在两个下游呈现为两种顺序（这正是并发下把帧重排的入口）。
-func TestEventTapFansOutOneFramePerSinkInOneOrder(t *testing.T) {
-	window := fanoutWindow{entered: make(chan struct{}), finished: make(chan struct{})}
-	var mu sync.Mutex
-	var tapped, requested []squad.ProcessEvent
-	tap := &eventTap{next: window, on: func(ev squad.ProcessEvent) {
-		mu.Lock()
-		tapped = append(tapped, ev)
-		seen := len(tapped)
-		mu.Unlock()
-		if seen == 2 {
-			// 不持有采集锁去等：让另一条支线真的能穿过扇出窗口（否则本测试
-			// 只会测到自己）。
-			close(window.entered)
-			select {
-			case <-window.finished:
-			case <-time.After(250 * time.Millisecond):
-			}
-		}
-	}}
-	_, err := tap.Execute(context.Background(), squad.ExecRequest{
-		OnEvent: func(ev squad.ProcessEvent) {
-			mu.Lock()
-			defer mu.Unlock()
-			requested = append(requested, ev)
-		},
-	})
-	if err != nil {
-		t.Fatal(err)
-	}
-	want := []squad.ProcessEvent{
-		{Type: "tool_execution_start", Tool: "grep", CallID: "call_ok"},
-		{Type: "tool_execution_end", Tool: "grep", CallID: "call_ok", Status: squad.ToolStatusOK, Text: "found"},
-		{Type: "tool_execution_start", Tool: "grep", CallID: "call_err"},
-		{Type: "tool_execution_end", Tool: "grep", CallID: "call_err",
-			Status: squad.ToolStatusError, ErrorReason: "grep: permission denied"},
-	}
-	assertArrivalOrder(t, "tap", tapped, want)
-	assertArrivalOrder(t, "request", requested, want)
-	assertPaired(t, "tap", tapped)
-	assertPaired(t, "request", requested)
-}
-
-func TestEventTapWithoutRequestHookStillTaps(t *testing.T) {
-	var mu sync.Mutex
-	var got []squad.ProcessEvent
-	tap := &eventTap{next: concurrentToolExecutor{}, on: func(ev squad.ProcessEvent) {
-		mu.Lock()
-		defer mu.Unlock()
-		got = append(got, ev)
-	}}
-	if _, err := tap.Execute(context.Background(), squad.ExecRequest{}); err != nil {
-		t.Fatal(err)
-	}
-	assertPaired(t, "tap-only", got)
-}

adapters/native/register_test.go

@@ -47,85 +47,3 @@ func TestBuildOptionsBashPermissionGating(t *testing.T) {
 		t.Fatal("missing model must fail-closed")
 	}
 }
-
-func TestConfigIdentityReflectsToolAndClient(t *testing.T) {
-	base := Config{
-		Client:       &llm.Client{BaseURL: "http://x", APIKey: "k"},
-		DefaultModel: "m1", MaxTurns: 5, Timeout: time.Minute,
-		Exec: &tools.ExecConfig{Bash: "/bin/bash"},
-	}
-	otherExec := base
-	otherExec.Exec = &tools.ExecConfig{Bash: "/bin/zsh"}
-	if base.Identity() == otherExec.Identity() {
-		t.Fatal("Exec tool-config change must change Config.Identity")
-	}
-	otherClient := base
-	otherClient.Client = &llm.Client{BaseURL: "http://y", APIKey: "k"}
-	if base.Identity() == otherClient.Identity() {
-		t.Fatal("client change must change Config.Identity")
-	}
-	same := base
-	if base.Identity() != same.Identity() {
-		t.Fatal("identical config must keep Identity stable")
-	}
-}
-
-func TestRegisterFullAllowDenyTakesEffectOnNextResolve(t *testing.T) {
-	r := squad.NewBackendResolver()
-	cfg := Config{
-		Client:       &llm.Client{BaseURL: "http://x", APIKey: "k"},
-		DefaultModel: "m1", MaxTurns: 1, Timeout: time.Second,
-		Exec: &tools.ExecConfig{Bash: "/bin/bash"},
-	}
-	RegisterFull(r, cfg)
-
-	allowRole := squad.RoleSpec{Role: "eng", Backend: "agent-native", Model: "m1", BashPermission: squad.BashPermissionAllow}
-	denyRole := squad.RoleSpec{Role: "eng", Backend: "agent-native", Model: "m1", BashPermission: squad.BashPermissionDeny}
-
-	eAllow, _, err := r.Resolve(allowRole, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	eDeny, _, err := r.Resolve(denyRole, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	if eAllow == eDeny {
-		t.Fatal("allow→deny must not reuse cached executor on next Resolve")
-	}
-	// Identity of role configs must diverge so cache keys differ.
-	if squad.ConfigIdentity(allowRole) == squad.ConfigIdentity(denyRole) {
-		t.Fatal("role ConfigIdentity must include BashPermission")
-	}
-}
-
-func TestRegisterFullToolConfigReplacementTakesEffect(t *testing.T) {
-	r := squad.NewBackendResolver()
-	cfg1 := Config{
-		Client:       &llm.Client{BaseURL: "http://x", APIKey: "k"},
-		DefaultModel: "m1", MaxTurns: 1, Timeout: time.Second,
-		Exec: &tools.ExecConfig{Bash: "/bin/bash"},
-	}
-	cfg2 := Config{
-		Client:       &llm.Client{BaseURL: "http://x", APIKey: "k"},
-		DefaultModel: "m1", MaxTurns: 1, Timeout: time.Second,
-		Exec: &tools.ExecConfig{Bash: "/bin/zsh"},
-	}
-	if cfg1.Identity() == cfg2.Identity() {
-		t.Fatal("precondition: tool configs must differ in Identity")
-	}
-	RegisterFull(r, cfg1)
-	role := squad.RoleSpec{Role: "eng", Backend: "agent-native", Model: "m1", BashPermission: squad.BashPermissionAllow}
-	e1, _, err := r.Resolve(role, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	RegisterFull(r, cfg2)
-	e2, _, err := r.Resolve(role, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	if e1 == e2 {
-		t.Fatal("RegisterFull tool-config replacement must invalidate cache")
-	}
-}

adapters/sqlite/queue.go

@@ -189,33 +189,8 @@ func (s *Store) PromoteDeferred(at time.Time) (int, error) {
 	return int(n), nil
 }
 
-// claimSQL is the whole claim: one atomic conditional UPDATE whose two
-// guarded counts bound the agent by its own max_concurrency and the runtime by
-// max_concurrency over every in-flight task of that runtime (an agent bound to
-// it, or a task this runtime already leased). A count that cannot be read
-// fails the statement, so the claim is refused instead of allowed.
-const claimSQL = `UPDATE task SET status='dispatched', runtime_id=?, lease_until=?, updated_at=?
-	WHERE id=? AND status='queued'
-	AND (SELECT COUNT(*) FROM task WHERE agent_id=? AND status IN ('dispatched','running')) < ?
-	AND (SELECT COUNT(*) FROM task t WHERE t.status IN ('dispatched','running')
-		AND (t.runtime_id=? OR EXISTS (SELECT 1 FROM agent a WHERE a.id=t.agent_id AND a.runtime_id=?))) < ?`
-
-// claimCapacitySQL counts the agent's own and the runtime's in-flight tasks in
-// one read, so the pre-check below matches the guards of claimSQL.
-const claimCapacitySQL = `SELECT
-	(SELECT COUNT(*) FROM task WHERE agent_id=? AND status IN ('dispatched','running')),
-	(SELECT COUNT(*) FROM task t WHERE t.status IN ('dispatched','running')
-		AND (t.runtime_id=? OR EXISTS (SELECT 1 FROM agent a WHERE a.id=t.agent_id AND a.runtime_id=?)))`
-
 // Claim takes one queued task whose agent and runtime are under their caps.
 // The task becomes dispatched with a five-minute lease.
-//
-// Both caps live inside the single conditional UPDATE, so concurrent claims
-// can overshoot neither the agent cap nor the runtime total: the agent count
-// is bounded by agent.max_concurrency and the runtime count (every in-flight
-// task whose agent belongs to this runtime, plus every task this runtime
-// already leased) by runtime.max_concurrency. A read error is a refusal, never
-// a silent allow: the candidate is skipped, and no task is handed out.
 func (s *Store) Claim(runtimeID string, at time.Time) (Task, bool, error) {
 	rt, err := s.getRuntime(runtimeID)
 	if err != nil {
@@ -248,20 +223,20 @@ func (s *Store) Claim(runtimeID string, at time.Time) (Task, bool, error) {
 		if agent.RuntimeID != "" && agent.RuntimeID != runtimeID {
 			continue
 		}
-		// Pre-check only: the UPDATE below re-counts both caps under the write
-		// lock. A count that cannot be read means unknown capacity, so refuse.
-		var agentBusy, runtimeBusy int
-		err = s.db.QueryRow(claimCapacitySQL, cand.agentID, runtimeID, runtimeID).Scan(&agentBusy, &runtimeBusy)
-		if err != nil {
-			return Task{}, false, fmt.Errorf("claim capacity count: %w", err)
+		cap := agent.MaxConcurrency
+		if rt.MaxConcurrency < cap {
+			cap = rt.MaxConcurrency
 		}
-		if agentBusy >= agent.MaxConcurrency || runtimeBusy >= rt.MaxConcurrency {
+		var running int
+		_ = s.db.QueryRow(`SELECT COUNT(*) FROM task WHERE agent_id=? AND status IN ('dispatched','running')`, cand.agentID).Scan(&running)
+		if running >= cap {
 			continue
 		}
 		lease := at.UTC().Add(5 * time.Minute).Format(time.RFC3339Nano)
-		res, err := s.db.Exec(claimSQL,
-			runtimeID, lease, now(), cand.id, cand.agentID, agent.MaxConcurrency,
-			runtimeID, runtimeID, rt.MaxConcurrency)
+		res, err := s.db.Exec(`UPDATE task SET status='dispatched', runtime_id=?, lease_until=?, updated_at=?
+			WHERE id=? AND status='queued'
+			AND (SELECT COUNT(*) FROM task WHERE agent_id=? AND status IN ('dispatched','running')) < ?`,
+			runtimeID, lease, now(), cand.id, cand.agentID, cap)
 		if err != nil {
 			return Task{}, false, err
 		}

adapters/sqlite/queue_test.go

@@ -1,325 +0,0 @@
-package platform
-
-import (
-	"path/filepath"
-	"sync"
-	"testing"
-	"time"
-)
-
-// openQueue opens a hermetic store in a temp dir and returns it with its path.
-func openQueue(t *testing.T) (*Store, string) {
-	t.Helper()
-	path := filepath.Join(t.TempDir(), "queue.db")
-	s, err := Open(path)
-	if err != nil {
-		t.Fatal(err)
-	}
-	t.Cleanup(func() { _ = s.Close() })
-	return s, path
-}
-
-// newQueueRuntime registers a runtime with the given cap.
-func newQueueRuntime(t *testing.T, s *Store, workspaceID, name string, cap int) Runtime {
-	t.Helper()
-	rt, err := s.RegisterRuntime(Runtime{
-		WorkspaceID: workspaceID, Name: name, Command: "in-process",
-		Protocol: "agent-native", MaxConcurrency: cap,
-	})
-	if err != nil {
-		t.Fatal(err)
-	}
-	return rt
-}
-
-// newQueueAgent creates an agent with the given cap, pinned to runtimeID
-// (an empty runtimeID leaves the agent unbound, claimable by any runtime).
-func newQueueAgent(t *testing.T, s *Store, workspaceID, name, runtimeID string, cap int) Agent {
-	t.Helper()
-	agent, err := s.CreateAgent(Agent{WorkspaceID: workspaceID, Name: name, RuntimeID: runtimeID, MaxConcurrency: cap})
-	if err != nil {
-		t.Fatal(err)
-	}
-	return agent
-}
-
-// queueTask creates an issue and queues one task for agentID.
-func queueTask(t *testing.T, s *Store, workspaceID, agentID, title string) Task {
-	t.Helper()
-	issue, err := s.CreateIssue(Issue{WorkspaceID: workspaceID, Title: title, Body: title, Status: "todo"})
-	if err != nil {
-		t.Fatal(err)
-	}
-	task, err := s.enqueue(issue, agentID, false, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	return task
-}
-
-// taskStatus reads the durable status of one task.
-func taskStatus(t *testing.T, s *Store, id string) string {
-	t.Helper()
-	var status string
-	if err := s.db.QueryRow(`SELECT status FROM task WHERE id=?`, id).Scan(&status); err != nil {
-		t.Fatal(err)
-	}
-	return status
-}
-
-// R6: the runtime cap bounds every agent on that runtime, not one agent only.
-// A per-agent cap of 4 each still yields a single in-flight task at cap 1.
-func TestClaimEnforcesRuntimeCapAcrossAgents(t *testing.T) {
-	for _, tc := range []struct {
-		name     string
-		bindToRT bool
-	}{
-		{name: "agents bound to the runtime", bindToRT: true},
-		{name: "agents not bound to the runtime", bindToRT: false},
-	} {
-		t.Run(tc.name, func(t *testing.T) {
-			s, _ := openQueue(t)
-			ws, err := s.CreateWorkspace("lab")
-			if err != nil {
-				t.Fatal(err)
-			}
-			rt := newQueueRuntime(t, s, ws.ID, "rt", 1)
-			runtimeID := ""
-			if tc.bindToRT {
-				runtimeID = rt.ID
-			}
-			first := newQueueAgent(t, s, ws.ID, "a1", runtimeID, 4)
-			second := newQueueAgent(t, s, ws.ID, "a2", runtimeID, 4)
-			t1 := queueTask(t, s, ws.ID, first.ID, "one")
-			t2 := queueTask(t, s, ws.ID, second.ID, "two")
-
-			claimed, ok, err := s.Claim(rt.ID, time.Now())
-			if err != nil || !ok || claimed.ID != t1.ID || claimed.RuntimeID != rt.ID {
-				t.Fatalf("first claim %+v ok=%v err=%v want task %s on %s", claimed, ok, err, t1.ID, rt.ID)
-			}
-			if _, ok, err := s.Claim(rt.ID, time.Now()); err != nil || ok {
-				t.Fatalf("runtime cap 1 must refuse the other agent: ok=%v err=%v", ok, err)
-			}
-			if got := taskStatus(t, s, t2.ID); got != "queued" {
-				t.Fatalf("second task %q must stay queued while the runtime is full", got)
-			}
-
-			if err := s.FinishTask(claimed.ID, "completed", ""); err != nil {
-				t.Fatal(err)
-			}
-			next, ok, err := s.Claim(rt.ID, time.Now())
-			if err != nil || !ok || next.ID != t2.ID {
-				t.Fatalf("claim after release %+v ok=%v err=%v want task %s", next, ok, err, t2.ID)
-			}
-		})
-	}
-}
-
-// The agent cap holds when two runtimes compete for the same agent.
-func TestClaimEnforcesAgentCapAcrossRuntimes(t *testing.T) {
-	s, _ := openQueue(t)
-	ws, err := s.CreateWorkspace("lab")
-	if err != nil {
-		t.Fatal(err)
-	}
-	rtA := newQueueRuntime(t, s, ws.ID, "rt-a", 2)
-	rtB := newQueueRuntime(t, s, ws.ID, "rt-b", 2)
-	agent := newQueueAgent(t, s, ws.ID, "solo", "", 1)
-	t1 := queueTask(t, s, ws.ID, agent.ID, "one")
-	t2 := queueTask(t, s, ws.ID, agent.ID, "two")
-
-	claimed, ok, err := s.Claim(rtA.ID, time.Now())
-	if err != nil || !ok || claimed.ID != t1.ID {
-		t.Fatalf("rt-a claim %+v ok=%v err=%v want task %s", claimed, ok, err, t1.ID)
-	}
-	if _, ok, err := s.Claim(rtB.ID, time.Now()); err != nil || ok {
-		t.Fatalf("agent cap 1 must bound the second runtime: ok=%v err=%v", ok, err)
-	}
-	if got := taskStatus(t, s, t2.ID); got != "queued" {
-		t.Fatalf("second task %q must stay queued while the agent is busy", got)
-	}
-}
-
-// Concurrent claims never overshoot the runtime cap or an agent cap.
-func TestClaimConcurrentNeverExceedsCaps(t *testing.T) {
-	s, _ := openQueue(t)
-	ws, err := s.CreateWorkspace("lab")
-	if err != nil {
-		t.Fatal(err)
-	}
-	rt := newQueueRuntime(t, s, ws.ID, "rt", 2)
-	tight := newQueueAgent(t, s, ws.ID, "tight", rt.ID, 1)
-	loose := newQueueAgent(t, s, ws.ID, "loose", rt.ID, 8)
-	other := newQueueAgent(t, s, ws.ID, "other", rt.ID, 8)
-	for _, agent := range []Agent{tight, loose, other} {
-		queueTask(t, s, ws.ID, agent.ID, "task for "+agent.Name)
-	}
-
-	const workers = 12
-	var wg sync.WaitGroup
-	var mu sync.Mutex
-	claimed := map[string]string{} // task id -> agent id
-	failures := []string{}
-	for i := 0; i < workers; i++ {
-		wg.Add(1)
-		go func() {
-			defer wg.Done()
-			task, ok, err := s.Claim(rt.ID, time.Now())
-			mu.Lock()
-			defer mu.Unlock()
-			switch {
-			case err != nil:
-				failures = append(failures, err.Error())
-			case ok:
-				claimed[task.ID] = task.AgentID
-			}
-		}()
-	}
-	wg.Wait()
-
-	if len(failures) > 0 {
-		t.Fatalf("concurrent claims failed: %v", failures)
-	}
-	if len(claimed) == 0 {
-		t.Fatal("no claim landed")
-	}
-	if len(claimed) > rt.MaxConcurrency {
-		t.Fatalf("runtime cap %d exceeded: %v", rt.MaxConcurrency, claimed)
-	}
-	perAgent := map[string]int{}
-	for _, agentID := range claimed {
-		perAgent[agentID]++
-	}
-	caps := map[string]int{tight.ID: tight.MaxConcurrency, loose.ID: loose.MaxConcurrency, other.ID: other.MaxConcurrency}
-	for agentID, n := range perAgent {
-		if cap, known := caps[agentID]; !known {
-			t.Fatalf("unknown agent %s claimed", agentID)
-		} else if n > cap {
-			t.Fatalf("agent cap %d exceeded by %d for %s: %v", cap, n, agentID, claimed)
-		}
-	}
-	if got := perAgent[tight.ID]; got > 1 {
-		t.Fatalf("agent cap 1 exceeded by %d: %v", got, claimed)
-	}
-}
-
-// A completed or lease-expired task gives its slot back.
-func TestClaimCapacityFreedByReleaseAndReclaim(t *testing.T) {
-	s, _ := openQueue(t)
-	ws, err := s.CreateWorkspace("lab")
-	if err != nil {
-		t.Fatal(err)
-	}
-	rt := newQueueRuntime(t, s, ws.ID, "rt", 1)
-	agent := newQueueAgent(t, s, ws.ID, "solo", rt.ID, 1)
-	t1 := queueTask(t, s, ws.ID, agent.ID, "one")
-	t2 := queueTask(t, s, ws.ID, agent.ID, "two")
-
-	claimed, ok, err := s.Claim(rt.ID, time.Now())
-	if err != nil || !ok || claimed.ID != t1.ID {
-		t.Fatalf("first claim %+v ok=%v err=%v want task %s", claimed, ok, err, t1.ID)
-	}
-	if _, ok, err := s.Claim(rt.ID, time.Now()); err != nil || ok {
-		t.Fatalf("the agent is at its cap: ok=%v err=%v", ok, err)
-	}
-
-	if err := s.FinishTask(claimed.ID, "completed", ""); err != nil {
-		t.Fatal(err)
-	}
-	second, ok, err := s.Claim(rt.ID, time.Now())
-	if err != nil || !ok || second.ID != t2.ID {
-		t.Fatalf("claim after release %+v ok=%v err=%v want task %s", second, ok, err, t2.ID)
-	}
-
-	n, err := s.RecoverOrphans(time.Now().Add(10 * time.Minute))
-	if err != nil || n != 1 {
-		t.Fatalf("reclaim %d %v", n, err)
-	}
-	if got := taskStatus(t, s, t2.ID); got != "queued" {
-		t.Fatalf("reclaimed task %q must be queued again", got)
-	}
-	third, ok, err := s.Claim(rt.ID, time.Now())
-	if err != nil || !ok || third.ID != t2.ID || third.Attempts != 1 {
-		t.Fatalf("claim after reclaim %+v ok=%v err=%v want task %s with one attempt", third, ok, err, t2.ID)
-	}
-}
-
-// A capacity read that fails refuses the claim; it never allows it.
-func TestClaimRefusesWhenReadFails(t *testing.T) {
-	t.Run("unreadable store", func(t *testing.T) {
-		s, path := openQueue(t)
-		ws, err := s.CreateWorkspace("lab")
-		if err != nil {
-			t.Fatal(err)
-		}
-		rt := newQueueRuntime(t, s, ws.ID, "rt", 2)
-		agent := newQueueAgent(t, s, ws.ID, "a", rt.ID, 2)
-		task := queueTask(t, s, ws.ID, agent.ID, "one")
-		if err := s.Close(); err != nil {
-			t.Fatal(err)
-		}
-		if claimed, ok, err := s.Claim(rt.ID, time.Now()); ok {
-			t.Fatalf("an unreadable store handed out %+v (err=%v)", claimed, err)
-		}
-		again, err := Open(path)
-		if err != nil {
-			t.Fatal(err)
-		}
-		defer again.Close()
-		if got := taskStatus(t, again, task.ID); got != "queued" {
-			t.Fatalf("task %q after a failed capacity read", got)
-		}
-	})
-
-	t.Run("agent lookup fails", func(t *testing.T) {
-		s, _ := openQueue(t)
-		ws, err := s.CreateWorkspace("lab")
-		if err != nil {
-			t.Fatal(err)
-		}
-		rt := newQueueRuntime(t, s, ws.ID, "rt", 2)
-		agent := newQueueAgent(t, s, ws.ID, "a", rt.ID, 2)
-		task := queueTask(t, s, ws.ID, agent.ID, "one")
-		if _, err := s.db.Exec(`DROP TABLE agent`); err != nil {
-			t.Fatal(err)
-		}
-		claimed, ok, err := s.Claim(rt.ID, time.Now())
-		if ok || err != nil {
-			t.Fatalf("a failed agent read must skip the candidate: %+v ok=%v err=%v", claimed, ok, err)
-		}
-		if _, err := s.db.Exec(`CREATE TABLE agent (id TEXT PRIMARY KEY, runtime_id TEXT)`); err != nil {
-			t.Fatal(err)
-		}
-		if got := taskStatus(t, s, task.ID); got != "queued" {
-			t.Fatalf("task %q after a skipped candidate", got)
-		}
-	})
-
-	// The guarded count runs inside the atomic claim: when it cannot be read
-	// the statement must fail rather than hand the task out.
-	t.Run("guard count fails", func(t *testing.T) {
-		s, _ := openQueue(t)
-		ws, err := s.CreateWorkspace("lab")
-		if err != nil {
-			t.Fatal(err)
-		}
-		rt := newQueueRuntime(t, s, ws.ID, "rt", 2)
-		agent := newQueueAgent(t, s, ws.ID, "a", rt.ID, 2)
-		task := queueTask(t, s, ws.ID, agent.ID, "one")
-		if _, err := s.db.Exec(`DROP TABLE agent`); err != nil {
-			t.Fatal(err)
-		}
-		lease := time.Now().UTC().Add(5 * time.Minute).Format(time.RFC3339Nano)
-		res, err := s.db.Exec(claimSQL, rt.ID, lease, now(), task.ID, agent.ID, 2, rt.ID, rt.ID, rt.MaxConcurrency)
-		if err == nil {
-			n, _ := res.RowsAffected()
-			t.Fatalf("a failed guard count claimed %d task(s)", n)
-		}
-		if _, err := s.db.Exec(`CREATE TABLE agent (id TEXT PRIMARY KEY, runtime_id TEXT)`); err != nil {
-			t.Fatal(err)
-		}
-		if got := taskStatus(t, s, task.ID); got != "queued" {
-			t.Fatalf("task %q after a failed guard count", got)
-		}
-	})
-}

adapters/sqlite/runstore.go

@@ -22,19 +22,13 @@ func (s *Store) RunStore() *SQLiteStore {
 	return &SQLiteStore{db: s.db}
 }
 
-// runSchemaVersion 是 squad run 存储的 schema 版本（squad_run_schema_version 记档）。
-// v1 起步，v2 补齐协作态列（mailbox/handoffs/board）。迁移只做增量：
-// 新库直接建到最新形态，旧库经 ALTER TABLE ADD COLUMN 补列（既有行读出为空默认，不丢数据）。
-const runSchemaVersion = 2
-
 const runSchema = `
 CREATE TABLE IF NOT EXISTS squad_run_state (
   ws_dir TEXT PRIMARY KEY,
   meta TEXT NOT NULL, workflow TEXT NOT NULL, squad TEXT NOT NULL, bb TEXT NOT NULL,
   template TEXT, approved TEXT, repair_budget_remaining INTEGER NOT NULL DEFAULT 0,
   repair_budget_granted INTEGER NOT NULL DEFAULT 0,
-  clarifications TEXT, steering TEXT, usage TEXT, updated_at TEXT NOT NULL,
-  mailbox TEXT, handoffs TEXT, board TEXT
+  clarifications TEXT, steering TEXT, usage TEXT, updated_at TEXT NOT NULL
 );
 CREATE TABLE IF NOT EXISTS squad_run_gate (
   ws_dir TEXT PRIMARY KEY, gate TEXT NOT NULL
@@ -47,90 +41,13 @@ CREATE TABLE IF NOT EXISTS squad_run_checkpoint (
 CREATE TABLE IF NOT EXISTS squad_run_leader (
   ws_dir TEXT PRIMARY KEY, data TEXT NOT NULL
 );
-CREATE TABLE IF NOT EXISTS squad_run_schema_version (
-  id INTEGER PRIMARY KEY CHECK (id = 1), version INTEGER NOT NULL
-);
 `
 
-// runStateV2Columns 是 v2 在 squad_run_state 上新增的协作态列（列名为代码内常量，非外部输入）。
-var runStateV2Columns = []string{"mailbox", "handoffs", "board"}
-
-// migrate 建表并按版本步进：老库补列后再记版本号，逐次读写都是幂等的
-// （每次 Save/Load 都调；补列只在版本落后时发生，且既有行不被改写）。
 func (s *SQLiteStore) migrate() error {
-	if _, err := s.db.Exec(runSchema); err != nil {
-		return err
-	}
-	cur, err := s.runSchemaVersion()
-	if err != nil {
-		return err
-	}
-	if cur >= runSchemaVersion {
-		return nil
-	}
-	if err := s.addRunStateColumns(runStateV2Columns); err != nil {
-		return err
-	}
-	_, err = s.db.Exec(`INSERT INTO squad_run_schema_version (id, version) VALUES (1, ?)
-	  ON CONFLICT(id) DO UPDATE SET version = excluded.version`, runSchemaVersion)
+	_, err := s.db.Exec(runSchema)
 	return err
 }
 
-// runSchemaVersion 读取当前 schema 版本；无记录 = 0（v1 及更早的库）。
-func (s *SQLiteStore) runSchemaVersion() (int, error) {
-	var v int
-	err := s.db.QueryRow(`SELECT version FROM squad_run_schema_version WHERE id = 1`).Scan(&v)
-	if err != nil {
-		if err == sql.ErrNoRows {
-			return 0, nil
-		}
-		return 0, err
-	}
-	return v, nil
-}
-
-// runStateColumns 读出 squad_run_state 现存列名（迁移判存在性用）。
-func (s *SQLiteStore) runStateColumns() (map[string]bool, error) {
-	rows, err := s.db.Query(`PRAGMA table_info(squad_run_state)`)
-	if err != nil {
-		return nil, err
-	}
-	defer rows.Close()
-	out := map[string]bool{}
-	for rows.Next() {
-		var cid, notnull, pk int
-		var name, ctype string
-		var dflt sql.NullString
-		if err := rows.Scan(&cid, &name, &ctype, &notnull, &dflt, &pk); err != nil {
-			return nil, err
-		}
-		out[name] = true
-	}
-	return out, rows.Err()
-}
-
-// addRunStateColumns 为缺失的列补 ALTER TABLE ADD COLUMN（既有行保持原值，
-// 新列读出为 NULL → 空默认）。
-func (s *SQLiteStore) addRunStateColumns(cols []string) error {
-	have, err := s.runStateColumns()
-	if err != nil {
-		return err
-	}
-	for _, col := range cols {
-		if have[col] {
-			continue
-		}
-		if _, err := s.db.Exec(`ALTER TABLE squad_run_state ADD COLUMN ` + col + ` TEXT`); err != nil {
-			// 并发迁移：另一进程可能刚补上同一列，复核后不算失败。
-			if again, err2 := s.runStateColumns(); err2 == nil && again[col] {
-				continue
-			}
-			return fmt.Errorf("migrate squad_run_state: add column %s: %w", col, err)
-		}
-	}
-	return nil
-}
-
 // OpenSQLiteStore 独立打开一个 squad.Store（不经 platform.Store 的场景，如测试）。
 func OpenSQLiteStore(path string) (*SQLiteStore, error) {
 	db, err := sql.Open("sqlite", path+"?_pragma=busy_timeout(5000)&_pragma=journal_mode(WAL)")
@@ -153,114 +70,54 @@ func j(v any) string {
 	return string(b)
 }
 
-// runStateColumn 是 squad_run_state 的一列：列名 + 待编码值（编码失败即报错，
-// 不用空串顶替——空串读回来是「空默认」，等于静默丢数据）。
-type runStateColumn struct {
-	name string
-	val  any
-}
-
-// runStateColumnsOf 列出 RunState 的全部持久化字段（含协作态 mailbox/handoffs/board）。
-func runStateColumnsOf(st *squad.RunState) []runStateColumn {
-	return []runStateColumn{
-		{"meta", st.Meta},
-		{"workflow", st.Plan.Workflow},
-		{"squad", st.Plan.Squad},
-		{"bb", st.BB},
-		{"approved", st.Approved},
-		{"clarifications", st.Clarifications},
-		{"steering", st.Steering},
-		{"usage", st.Usage},
-		{"mailbox", st.Mailbox},
-		{"handoffs", st.Handoffs},
-		{"board", st.Board},
-	}
-}
-
 func (s *SQLiteStore) SaveRun(st *squad.RunState) error {
 	if err := s.migrate(); err != nil {
 		return err
 	}
-	encoded := map[string]string{}
-	for _, col := range runStateColumnsOf(st) {
-		data, err := json.Marshal(col.val)
-		if err != nil {
-			return fmt.Errorf("save run %s: encode column %s: %w", st.Meta.WorkDir, col.name, err)
-		}
-		encoded[col.name] = string(data)
-	}
 	_, err := s.db.Exec(`INSERT INTO squad_run_state
-	  (ws_dir, meta, workflow, squad, bb, template, approved, repair_budget_remaining, repair_budget_granted, clarifications, steering, usage, mailbox, handoffs, board, updated_at)
-	  VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)
+	  (ws_dir, meta, workflow, squad, bb, template, approved, repair_budget_remaining, repair_budget_granted, clarifications, steering, usage, updated_at)
+	  VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?)
 	  ON CONFLICT(ws_dir) DO UPDATE SET
 	    meta=excluded.meta, workflow=excluded.workflow, squad=excluded.squad, bb=excluded.bb,
 	    template=excluded.template, approved=excluded.approved,
 	    repair_budget_remaining=excluded.repair_budget_remaining,
 	    repair_budget_granted=excluded.repair_budget_granted,
 	    clarifications=excluded.clarifications, steering=excluded.steering,
-	    usage=excluded.usage, mailbox=excluded.mailbox, handoffs=excluded.handoffs,
-	    board=excluded.board, updated_at=excluded.updated_at`,
-		st.Meta.WorkDir, encoded["meta"], encoded["workflow"], encoded["squad"], encoded["bb"],
-		st.Template, encoded["approved"], st.RepairBudgetRemaining, st.RepairBudgetGranted,
-		encoded["clarifications"], encoded["steering"], encoded["usage"],
-		encoded["mailbox"], encoded["handoffs"], encoded["board"], now())
+	    usage=excluded.usage, updated_at=excluded.updated_at`,
+		st.Meta.WorkDir, j(st.Meta), j(st.Plan.Workflow), j(st.Plan.Squad), j(st.BB),
+		st.Template, j(st.Approved), st.RepairBudgetRemaining, st.RepairBudgetGranted,
+		j(st.Clarifications), j(st.Steering), j(st.Usage), now())
 	return err
 }
 
-// LoadRun 按列位置读回 RunState。解码目标逐列具名指定——绝不用 JSON 文本当键：
-// 文本相同的两列（例如两条一模一样的澄清/补充指令）会互相覆盖，把一列丢掉。
 func (s *SQLiteStore) LoadRun(wsDir string) (*squad.RunState, error) {
 	if err := s.migrate(); err != nil {
 		return nil, err
 	}
 	st := &squad.RunState{}
-	// 列 → 扫描目标（顺序与 SQL 一一对应）。
-	var tpl sql.NullString
-	var granted int
-	var meta, wf, sq, bb, approved, clar, steering, usage, mailbox, handoffs, board sql.NullString
+	var meta, wf, sq, bb, template, approved, clar, steering, usage string
 	row := s.db.QueryRow(`SELECT meta, workflow, squad, bb, template, approved,
-	  repair_budget_remaining, repair_budget_granted, clarifications, steering, usage,
-	  mailbox, handoffs, board
+	  repair_budget_remaining, repair_budget_granted, clarifications, steering, usage
 	  FROM squad_run_state WHERE ws_dir = ?`, wsDir)
-	if err := row.Scan(&meta, &wf, &sq, &bb, &tpl, &approved,
-		&st.RepairBudgetRemaining, &granted, &clar, &steering, &usage,
-		&mailbox, &handoffs, &board); err != nil {
+	var granted int
+	err := row.Scan(&meta, &wf, &sq, &bb, &template, &approved,
+		&st.RepairBudgetRemaining, &granted, &clar, &steering, &usage)
+	if err != nil {
 		if err == sql.ErrNoRows {
 			return nil, fmt.Errorf("load run %s: %w", wsDir, os.ErrNotExist)
 		}
 		return nil, err
 	}
 	st.RepairBudgetGranted = granted
-	for _, dec := range []struct {
-		col string
-		raw sql.NullString
-		dst any
-	}{
-		{"meta", meta, &st.Meta},
-		{"workflow", wf, &st.Plan.Workflow},
-		{"squad", sq, &st.Plan.Squad},
-		{"bb", bb, &st.BB},
-		{"approved", approved, &st.Approved},
-		{"clarifications", clar, &st.Clarifications},
-		{"steering", steering, &st.Steering},
-		{"usage", usage, &st.Usage},
-		{"mailbox", mailbox, &st.Mailbox},
-		{"handoffs", handoffs, &st.Handoffs},
-		{"board", board, &st.Board},
+	for pair, dst := range map[string]any{
+		meta: &st.Meta, wf: &st.Plan.Workflow, sq: &st.Plan.Squad, bb: &st.BB,
+		approved: &st.Approved, clar: &st.Clarifications, steering: &st.Steering, usage: &st.Usage,
 	} {
-		// 旧行缺列（NULL）或空串 = 该字段的空默认，不报错、不影响其他列。
-		if !dec.raw.Valid || dec.raw.String == "" {
-			continue
-		}
-		if err := json.Unmarshal([]byte(dec.raw.String), dec.dst); err != nil {
-			return nil, fmt.Errorf("load run %s: parse column %s: %w", wsDir, dec.col, err)
+		if err := json.Unmarshal([]byte(pair), dst); err != nil {
+			return nil, fmt.Errorf("load run %s: parse: %w", wsDir, err)
 		}
 	}
-	st.Template = tpl.String
-	if st.Meta.WorkDir == "" {
-		// 行键（ws_dir）即权威工作区路径：meta 旧数据缺该字段时由行键兜底。
-		st.Meta.WorkDir = wsDir
-	}
+	st.Template = template
 	if st.Approved == nil {
 		st.Approved = map[string]bool{}
 	}

adapters/sqlite/runstore_test.go

@@ -2,14 +2,8 @@ package platform
 
 import (
 	"context"
-	"database/sql"
-	"encoding/json"
-	"fmt"
 	"os"
 	"path/filepath"
-	"reflect"
-	"strings"
-	"sync"
 	"testing"
 	"time"
 
@@ -17,12 +11,10 @@ import (
 )
 
 func newTestRunState(wsDir string) *squad.RunState {
-	// 固定时刻：多次构造出的状态除工作区路径外完全一致，可直接比对两个后端的读回结果。
-	now := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
 	return &squad.RunState{
 		Meta: squad.RunMetadata{
 			ID: "run_test", Status: "executing", WorkDir: wsDir,
-			CreatedAt: now, UpdatedAt: now,
+			CreatedAt: time.Now(), UpdatedAt: time.Now(),
 			Metadata: map[string]string{"mode": ""},
 		},
 		Plan: squad.SquadPlan{
@@ -35,28 +27,14 @@ func newTestRunState(wsDir string) *squad.RunState {
 			},
 			Squad: squad.SquadConfig{SquadName: "demo"},
 		},
-		BB:                    squad.BlackboardState{Facts: []squad.FactItem{{ID: "f1", Claim: "x"}}, Fields: map[string]string{"topic": "demo"}},
+		BB:                    squad.BlackboardState{Facts: []squad.FactItem{{ID: "f1", Claim: "x"}}},
 		Template:              "tpl",
 		Approved:              map[string]bool{"step:s1": true},
 		RepairBudgetRemaining: 1,
 		RepairBudgetGranted:   0,
-		// 内容完全相同的重复条目（且 clarifications 与 steering 的 JSON 文本一样）：
-		// 内容键寻址的解码会把它们互相覆盖掉——这里逐条都要活下来。
-		Clarifications: []string{"同一句澄清", "同一句澄清"},
-		Steering:       []string{"同一句澄清", "同一句澄清"},
-		Mailbox: []squad.MailMessage{
-			{ID: "m1", From: "researcher", To: "writer", Text: "同一句澄清", At: now},
-			{ID: "m2", From: "researcher", To: "writer", Text: "同一句澄清", At: now},
-		},
-		Handoffs: []squad.HandoffRecord{
-			{FromRole: "researcher", FromStep: "s1", ToRole: "writer", Output: "同一句澄清", At: now},
-			{FromRole: "writer", FromStep: "s2", ToRole: "reviewer", Output: "同一句澄清", At: now},
-		},
-		Board: []squad.BoardTask{
-			{ID: "s1", Status: "done", Attempts: 1},
-			{ID: "s2", Status: "claimed", ClaimedBy: "writer", LeaseExpires: now.Add(time.Minute), Attempts: 2},
-		},
-		Usage: map[string]*squad.Usage{"writer": {PromptTokens: 10, CompletionTokens: 5, Steps: 1}},
+		Clarifications:        []string{"c1"},
+		Steering:              []string{"st1"},
+		Usage:                 map[string]*squad.Usage{"writer": {PromptTokens: 10, CompletionTokens: 5, Steps: 1}},
 	}
 }
 
@@ -79,20 +57,9 @@ func TestSQLiteStoreRoundTrip(t *testing.T) {
 	if got.Meta.ID != "run_test" || got.Plan.Workflow.WorkflowID != "wf1" || got.Template != "tpl" {
 		t.Fatalf("loaded = %+v", got.Meta)
 	}
-	if !got.Approved["step:s1"] || got.RepairBudgetRemaining != 1 ||
-		len(got.Clarifications) != 2 || len(got.Steering) != 2 {
+	if !got.Approved["step:s1"] || got.RepairBudgetRemaining != 1 || len(got.Clarifications) != 1 || len(got.Steering) != 1 {
 		t.Fatalf("scalars = %+v", got)
 	}
-	// 协作态必须真的落库（F11a）：重启后邮箱/移交/任务板不能静默丢。
-	if len(got.Mailbox) != 2 || got.Mailbox[0].ID != "m1" || got.Mailbox[1].ID != "m2" {
-		t.Fatalf("mailbox = %+v", got.Mailbox)
-	}
-	if len(got.Handoffs) != 2 || got.Handoffs[1].ToRole != "reviewer" {
-		t.Fatalf("handoffs = %+v", got.Handoffs)
-	}
-	if len(got.Board) != 2 || got.Board[1].Status != "claimed" || got.Board[1].ClaimedBy != "writer" {
-		t.Fatalf("board = %+v", got.Board)
-	}
 	if got.Usage["writer"].PromptTokens != 10 {
 		t.Fatalf("usage = %+v", got.Usage)
 	}
@@ -160,7 +127,7 @@ func TestEngineRunsOnSQLiteStore(t *testing.T) {
 		}
 	}
 	backends := squad.NewBackendResolver()
-	backends.RegisterInstance("script", stubExecutor{result: &squad.ExecResult{Success: true, Logs: "ok", Patches: map[string][]byte{"output.md": []byte(goodReportLite), "notes.md": []byte(goodReportLite), "fields.json": []byte("{}")}}})
+	backends.RegisterInstance("script", stubExecutor{result: &squad.ExecResult{Success: false, Logs: "ok", Patches: map[string][]byte{"output.md": []byte(goodReportLite), "notes.md": []byte(goodReportLite), "fields.json": []byte("{}")}}})
 	blueprints := squad.NewBlueprintRegistry()
 	if err := blueprints.Register(squad.Blueprint{
 		Key: "demo", Title: "入库演示", Match: []string{"入库演示"},
@@ -218,542 +185,4 @@ func TestEngineRunsOnSQLiteStore(t *testing.T) {
 	}
 }
 
-// TestEngineFailedResultWithCompletePatchesStaysFailedOnSQLiteStore：F05/U3B 的
-// 不变式在 SQLite-store 集成层同样成立——executor 报失败就是失败，patches
-// （交付物）齐全也不得把运行翻成 completed：文件只是诊断信息，不是权威结果。
-func TestEngineFailedResultWithCompletePatchesStaysFailedOnSQLiteStore(t *testing.T) {
-	runStore, err := OpenSQLiteStore(filepath.Join(t.TempDir(), "runs.db"))
-	if err != nil {
-		t.Fatal(err)
-	}
-	defer runStore.Close()
-	dir := t.TempDir()
-	reg := squad.NewRoleRegistry()
-	if err := reg.Register(squad.RoleSpec{Role: "researcher", Backend: "script"}); err != nil {
-		t.Fatal(err)
-	}
-	backends := squad.NewBackendResolver()
-	// 失败结果但携带全套交付物：旧语义会把它提升为成功，U3B 起必须保持失败。
-	backends.RegisterInstance("script", stubExecutor{result: &squad.ExecResult{
-		Success: false, Logs: "stub-refused",
-		Patches: map[string][]byte{
-			"output.md":   []byte(goodReportLite),
-			"notes.md":    []byte(goodReportLite),
-			"fields.json": []byte("{}"),
-		},
-	}})
-	blueprints := squad.NewBlueprintRegistry()
-	if err := blueprints.Register(squad.Blueprint{
-		Key: "faildemo", Title: "失败演示", Match: []string{"失败演示"},
-		Steps: []squad.StepSpec{{ID: "s1", Role: "researcher", Title: "调研", Targets: []string{"output.md"}}},
-	}); err != nil {
-		t.Fatal(err)
-	}
-	eng, err := squad.NewEngine(squad.Options{
-		DataDir: dir, Registry: reg, Blueprints: blueprints,
-		Planner:  squad.NewRulePlanner(),
-		Backends: backends, Sink: squad.NopSink{},
-		Store: runStore,
-	})
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, err := eng.Run(context.Background(), squad.Goal{Text: "做一个失败演示", HitlMode: squad.HitlAuto}, squad.RunOptions{})
-	if err != nil {
-		t.Fatalf("run: %v", err)
-	}
-	if res.Status == squad.RunCompleted {
-		t.Fatalf("failed executor result must not complete (patches are diagnostic only)")
-	}
-	if res.Status != squad.RunFailed {
-		t.Fatalf("status = %s, want failed", res.Status)
-	}
-	if res.Plan == nil || res.Plan.Steps[0].Status != "failed" {
-		t.Fatalf("s1 should be failed, plan = %+v", res.Plan)
-	}
-	if !strings.Contains(res.Error, "stub-refused") {
-		t.Fatalf("failure reason not preserved: %q", res.Error)
-	}
-	// 同一断言穿过 Store 端口：SQLite 里落盘的也是失败，而非 completed。
-	st, err := runStore.LoadRun(res.WorkDir)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if st.Plan.Workflow.Status != "failed" {
-		t.Fatalf("persisted workflow status = %s, want failed", st.Plan.Workflow.Status)
-	}
-	if got := st.Plan.Workflow.Steps[0]; got.Status != "failed" || !strings.Contains(got.Error, "stub-refused") {
-		t.Fatalf("persisted step = %+v", got)
-	}
-	// patches 确已物化（文件齐全），依然不翻转状态：权威是执行结果，不是文件。
-	sub := filepath.FromSlash(st.Plan.Workflow.Steps[0].SubWorkspace)
-	if _, err := os.Stat(filepath.Join(res.WorkDir, sub, "output.md")); err != nil {
-		t.Fatalf("diagnostic patch should still be materialized: %v", err)
-	}
-}
-
 const goodReportLite = "# 测试交付物\n\n## 概述\n\n这是由脚本后端生成的合格交付物，长度超过五十个字符以满足最小质检要求。\n"
-
-// ---- F11a/F11b：SQLite 与 FileStore 同一契约、重复条目不丢、旧库升级 ----
-
-// assertSameRunState 逐字段比对两个后端读回的运行状态（同一契约）。
-// 字段按规范 JSON 比对：时间等值字段（创建/更新时间、消息时刻、租约到期）在两侧
-// 都经历过 JSON 往返，按墙钟等价判定，不受内部表示（时区指针/单调时钟）影响。
-func assertSameRunState(t *testing.T, got, want *squad.RunState) {
-	t.Helper()
-	sameJSON := func(label string, a, b any) {
-		t.Helper()
-		ab, err := json.Marshal(a)
-		if err != nil {
-			t.Fatalf("%s: marshal got: %v", label, err)
-		}
-		bb, err := json.Marshal(b)
-		if err != nil {
-			t.Fatalf("%s: marshal want: %v", label, err)
-		}
-		if string(ab) != string(bb) {
-			t.Fatalf("%s differs:\n got %s\nwant %s", label, ab, bb)
-		}
-	}
-	sameJSON("meta", got.Meta, want.Meta)
-	sameJSON("plan", got.Plan, want.Plan)
-	sameJSON("blackboard", got.BB, want.BB)
-	sameJSON("template", got.Template, want.Template)
-	sameJSON("approved", got.Approved, want.Approved)
-	sameJSON("repair_budget_remaining", got.RepairBudgetRemaining, want.RepairBudgetRemaining)
-	sameJSON("repair_budget_granted", got.RepairBudgetGranted, want.RepairBudgetGranted)
-	sameJSON("clarifications", got.Clarifications, want.Clarifications)
-	sameJSON("steering", got.Steering, want.Steering)
-	sameJSON("mailbox", got.Mailbox, want.Mailbox)
-	sameJSON("handoffs", got.Handoffs, want.Handoffs)
-	sameJSON("board", got.Board, want.Board)
-	sameJSON("usage", got.Usage, want.Usage)
-}
-
-// TestRunStoreContractParity：FileStore 与 SQLiteStore 满足同一份 Store 契约——
-// 同一份运行状态（含 mailbox/handoffs/board）写进去，读回来必须一致。
-func TestRunStoreContractParity(t *testing.T) {
-	sq, err := OpenSQLiteStore(filepath.Join(t.TempDir(), "runs.db"))
-	if err != nil {
-		t.Fatal(err)
-	}
-	defer sq.Close()
-
-	fileWS := filepath.Join(t.TempDir(), "runs", "run_file")
-	sqlWS := filepath.Join(t.TempDir(), "runs", "run_sql")
-	// FileStore 侧的模板文件由工作区装配期写出（PrepareWorkspace），先补齐，
-	// 使两侧的 SaveRun/LoadRun 覆盖同一批字段。
-	if err := os.MkdirAll(fileWS, 0o755); err != nil {
-		t.Fatal(err)
-	}
-	if err := os.WriteFile(filepath.Join(fileWS, "template.md"), []byte("tpl"), 0o644); err != nil {
-		t.Fatal(err)
-	}
-
-	want := newTestRunState(fileWS)
-	if err := (squad.FileStore{}).SaveRun(want); err != nil {
-		t.Fatal(err)
-	}
-	gotFile, err := (squad.FileStore{}).LoadRun(fileWS)
-	if err != nil {
-		t.Fatal(err)
-	}
-	sqlState := newTestRunState(sqlWS)
-	if err := sq.SaveRun(sqlState); err != nil {
-		t.Fatal(err)
-	}
-	gotSQL, err := sq.LoadRun(sqlWS)
-	if err != nil {
-		t.Fatal(err)
-	}
-
-	// SQLite 读回的内容与写进去的一致。
-	assertSameRunState(t, gotSQL, sqlState)
-	// 两个后端的读回结果逐字段一致（除工作区路径这一运行自身的键）。
-	assertSameRunState(t, gotFile, want)
-	fileMeta, sqlMeta := gotFile.Meta, gotSQL.Meta
-	fileMeta.WorkDir, sqlMeta.WorkDir = "", ""
-	fileJSON, _ := json.Marshal(fileMeta)
-	sqlJSON, _ := json.Marshal(sqlMeta)
-	if string(fileJSON) != string(sqlJSON) {
-		t.Fatalf("meta parity broken:\n file %s\n  sql %s", fileJSON, sqlJSON)
-	}
-}
-
-// TestSQLiteRunStoreKeepsIdenticalEntries：内容相同的两条澄清/补充指令必须都活着
-// （旧实现拿 JSON 文本当键，两条同文条目只剩一条）。
-func TestSQLiteRunStoreKeepsIdenticalEntries(t *testing.T) {
-	sq, err := OpenSQLiteStore(filepath.Join(t.TempDir(), "runs.db"))
-	if err != nil {
-		t.Fatal(err)
-	}
-	defer sq.Close()
-	ws := filepath.Join(t.TempDir(), "runs", "run_dup")
-	st := newTestRunState(ws)
-	if err := sq.SaveRun(st); err != nil {
-		t.Fatal(err)
-	}
-	got, err := sq.LoadRun(ws)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if !reflect.DeepEqual(got.Clarifications, []string{"同一句澄清", "同一句澄清"}) {
-		t.Fatalf("clarifications = %#v, want both identical entries", got.Clarifications)
-	}
-	if !reflect.DeepEqual(got.Steering, []string{"同一句澄清", "同一句澄清"}) {
-		t.Fatalf("steering = %#v, want both identical entries", got.Steering)
-	}
-	// 邮箱正文相同、ID 不同：两条都要在（不靠正文寻址）。
-	if len(got.Mailbox) != 2 || got.Mailbox[0].ID != "m1" || got.Mailbox[1].ID != "m2" {
-		t.Fatalf("mailbox = %+v", got.Mailbox)
-	}
-}
-
-// legacyRunStateDDL 是 F11 修复前的 squad_run_state（无协作态列），用于验证升级路径。
-const legacyRunStateDDL = `
-CREATE TABLE squad_run_state (
-  ws_dir TEXT PRIMARY KEY,
-  meta TEXT NOT NULL, workflow TEXT NOT NULL, squad TEXT NOT NULL, bb TEXT NOT NULL,
-  template TEXT, approved TEXT, repair_budget_remaining INTEGER NOT NULL DEFAULT 0,
-  repair_budget_granted INTEGER NOT NULL DEFAULT 0,
-  clarifications TEXT, steering TEXT, usage TEXT, updated_at TEXT NOT NULL
-);
-`
-
-// TestSQLiteRunStoreUpgradesLegacySchema：旧库（缺 mailbox/handoffs/board 列）打开后
-// 补列升级，既有一行原样读出，协作态为空默认不报错、不丢其他字段。
-func TestSQLiteRunStoreUpgradesLegacySchema(t *testing.T) {
-	path := filepath.Join(t.TempDir(), "runs.db")
-	legacy, err := sql.Open("sqlite", path)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if _, err := legacy.Exec(legacyRunStateDDL); err != nil {
-		t.Fatal(err)
-	}
-	ws := filepath.Join(t.TempDir(), "runs", "run_legacy")
-	meta, _ := json.Marshal(squad.RunMetadata{ID: "run_legacy", WorkDir: ws, Status: "executing"})
-	wf, _ := json.Marshal(squad.WorkflowPlan{WorkflowID: "wf_old", Status: "in_progress"})
-	sqCfg, _ := json.Marshal(squad.SquadConfig{SquadName: "legacy"})
-	bb, _ := json.Marshal(squad.BlackboardState{Fields: map[string]string{"k": "v"}})
-	approved, _ := json.Marshal(map[string]bool{"step:s1": true})
-	clar, _ := json.Marshal([]string{"旧澄清", "旧澄清"})
-	usage, _ := json.Marshal(map[string]*squad.Usage{"writer": {PromptTokens: 7, CompletionTokens: 3, Steps: 2}})
-	if _, err := legacy.Exec(`INSERT INTO squad_run_state
-	  (ws_dir, meta, workflow, squad, bb, template, approved, repair_budget_remaining,
-	   repair_budget_granted, clarifications, steering, usage, updated_at)
-	  VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?)`,
-		ws, meta, wf, sqCfg, bb, "旧模板", approved, 2, 1, clar, `["旧补充"]`, usage, "2026-01-01T00:00:00Z"); err != nil {
-		t.Fatal(err)
-	}
-	if err := legacy.Close(); err != nil {
-		t.Fatal(err)
-	}
-
-	sq, err := OpenSQLiteStore(path)
-	if err != nil {
-		t.Fatalf("open legacy db: %v", err)
-	}
-	defer sq.Close()
-
-	got, err := sq.LoadRun(ws)
-	if err != nil {
-		t.Fatalf("load upgraded row: %v", err)
-	}
-	// 旧数据无损。
-	if got.Meta.ID != "run_legacy" || got.Plan.Workflow.WorkflowID != "wf_old" ||
-		got.Plan.Squad.SquadName != "legacy" || got.Template != "旧模板" {
-		t.Fatalf("legacy row corrupted: %+v", got)
-	}
-	if !got.Approved["step:s1"] || got.RepairBudgetRemaining != 2 || got.RepairBudgetGranted != 1 {
-		t.Fatalf("legacy scalars lost: %+v", got)
-	}
-	if !reflect.DeepEqual(got.Clarifications, []string{"旧澄清", "旧澄清"}) {
-		t.Fatalf("legacy clarifications lost: %#v", got.Clarifications)
-	}
-	if !reflect.DeepEqual(got.Steering, []string{"旧补充"}) {
-		t.Fatalf("legacy steering lost: %#v", got.Steering)
-	}
-	if got.Usage["writer"].PromptTokens != 7 || got.BB.Fields["k"] != "v" {
-		t.Fatalf("legacy usage/blackboard lost: %+v %+v", got.Usage, got.BB)
-	}
-	// 缺列读出为空默认。
-	if len(got.Mailbox) != 0 || len(got.Handoffs) != 0 || len(got.Board) != 0 {
-		t.Fatalf("new columns should default to empty: %+v %+v %+v", got.Mailbox, got.Handoffs, got.Board)
-	}
-	// 版本已记档，补列只做一次。
-	var version int
-	if err := sq.db.QueryRow(`SELECT version FROM squad_run_schema_version WHERE id = 1`).Scan(&version); err != nil {
-		t.Fatalf("schema version not recorded: %v", err)
-	}
-	if version != runSchemaVersion {
-		t.Fatalf("schema version = %d, want %d", version, runSchemaVersion)
-	}
-	// 升级后写入协作态并再次读回。
-	got.Mailbox = []squad.MailMessage{{ID: "m1", From: "a", To: "b", Text: "升级后写入"}}
-	got.Board = []squad.BoardTask{{ID: "s1", Status: "done"}}
-	got.Meta.Status = "completed"
-	if err := sq.SaveRun(got); err != nil {
-		t.Fatalf("save after upgrade: %v", err)
-	}
-	again, err := sq.LoadRun(ws)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if len(again.Mailbox) != 1 || again.Mailbox[0].Text != "升级后写入" ||
-		len(again.Board) != 1 || again.Board[0].ID != "s1" {
-		t.Fatalf("post-upgrade collaboration state lost: %+v %+v", again.Mailbox, again.Board)
-	}
-	if !reflect.DeepEqual(again.Clarifications, []string{"旧澄清", "旧澄清"}) {
-		t.Fatalf("upgrade write clobbered legacy clarifications: %#v", again.Clarifications)
-	}
-}
-
-// TestSQLiteStoreKeepsCollaborationStateAcrossRestart：进程重启（新连接）后，
-// 邮箱消息、所有权移交记录与任务板状态都还在——恢复不必从头重建协作上下文。
-func TestSQLiteStoreKeepsCollaborationStateAcrossRestart(t *testing.T) {
-	dbPath := filepath.Join(t.TempDir(), "runs.db")
-	runStore, err := OpenSQLiteStore(dbPath)
-	if err != nil {
-		t.Fatal(err)
-	}
-	defer runStore.Close()
-
-	dir := t.TempDir()
-	reg := squad.NewRoleRegistry()
-	for _, role := range []string{"researcher", "writer", "reviewer"} {
-		if err := reg.Register(squad.RoleSpec{Role: role, Backend: "script"}); err != nil {
-			t.Fatal(err)
-		}
-	}
-	backends := squad.NewBackendResolver()
-	backends.RegisterInstance("script", stubExecutor{result: &squad.ExecResult{
-		Success: true, Logs: "ok",
-		Patches: map[string][]byte{"output.md": []byte(goodReportLite), "notes.md": []byte(goodReportLite), "fields.json": []byte("{}")},
-	}})
-	blueprints := squad.NewBlueprintRegistry()
-	if err := blueprints.Register(squad.Blueprint{
-		Key: "handoff", Title: "移交演示", Match: []string{"移交演示"},
-		Steps: []squad.StepSpec{
-			{ID: "s1", Role: "researcher", Title: "调研", Targets: []string{"notes.md", "fields.json"}},
-			{ID: "s2", Role: "writer", Title: "成文", Targets: []string{"output.md"}, Handoff: "reviewer"},
-			{ID: "s3", Role: "reviewer", Title: "终审", Targets: []string{"output.md"}},
-		},
-	}); err != nil {
-		t.Fatal(err)
-	}
-	if err := blueprints.Register(squad.Blueprint{
-		Key: "board", Title: "任务板演示", Match: []string{"任务板演示"},
-		Steps: []squad.StepSpec{
-			{ID: "a", Role: "researcher", Title: "任务a", Targets: []string{"notes.md"}},
-			{ID: "b", Role: "writer", Title: "任务b", Needs: []string{"a"}, Targets: []string{"output.md"}},
-		},
-	}); err != nil {
-		t.Fatal(err)
-	}
-	eng, err := squad.NewEngine(squad.Options{
-		DataDir: dir, Registry: reg, Blueprints: blueprints,
-		Planner:  squad.NewRulePlanner(),
-		Backends: backends, Sink: squad.NopSink{},
-		Store: runStore,
-	})
-	if err != nil {
-		t.Fatal(err)
-	}
-
-	ctx := context.Background()
-	handed, err := eng.Run(ctx, squad.Goal{Text: "做一个移交演示", HitlMode: squad.HitlAuto}, squad.RunOptions{})
-	if err != nil || handed.Status != squad.RunCompleted {
-		t.Fatalf("handoff run: %v %+v", err, handed)
-	}
-	if err := eng.PostMessage(ctx, handed.RunID, squad.MailMessage{From: "human", To: "writer", Text: "补充口径"}); err != nil {
-		t.Fatalf("post message: %v", err)
-	}
-	boarded, err := eng.Run(ctx, squad.Goal{Text: "做一个任务板演示", HitlMode: squad.HitlAuto},
-		squad.RunOptions{Topology: squad.TopologyTaskBoard})
-	if err != nil || boarded.Status != squad.RunCompleted {
-		t.Fatalf("board run: %v %+v", err, boarded)
-	}
-
-	// 重启 = 新连接（新进程会重新 migrate）。
-	restarted, err := OpenSQLiteStore(dbPath)
-	if err != nil {
-		t.Fatal(err)
-	}
-	defer restarted.Close()
-
-	gotHanded, err := restarted.LoadRun(handed.WorkDir)
-	if err != nil {
-		t.Fatalf("reload handoff run: %v", err)
-	}
-	if len(gotHanded.Handoffs) != 1 || gotHanded.Handoffs[0].FromRole != "writer" || gotHanded.Handoffs[0].ToRole != "reviewer" {
-		t.Fatalf("handoffs lost across restart: %+v", gotHanded.Handoffs)
-	}
-	if len(gotHanded.Mailbox) != 1 || gotHanded.Mailbox[0].Text != "补充口径" || gotHanded.Mailbox[0].To != "writer" {
-		t.Fatalf("mailbox lost across restart: %+v", gotHanded.Mailbox)
-	}
-	gotBoarded, err := restarted.LoadRun(boarded.WorkDir)
-	if err != nil {
-		t.Fatalf("reload board run: %v", err)
-	}
-	if len(gotBoarded.Board) != 2 {
-		t.Fatalf("board lost across restart: %+v", gotBoarded.Board)
-	}
-	for _, task := range gotBoarded.Board {
-		if task.Status != "done" {
-			t.Fatalf("board task %s = %s, want done", task.ID, task.Status)
-		}
-	}
-}
-
-// faultingStore 在真实 SQLiteStore 之上注入落盘故障（其余方法透传）。
-type faultingStore struct {
-	inner          squad.Store
-	failCheckpoint bool
-	failTerminal   bool
-}
-
-func (f *faultingStore) SaveRun(st *squad.RunState) error {
-	if f.failTerminal && squad.RunStatus(st.Meta.Status).Terminal() {
-		return fmt.Errorf("injected sqlite run state save failure (%s)", st.Meta.Status)
-	}
-	return f.inner.SaveRun(st)
-}
-
-func (f *faultingStore) LoadRun(wsDir string) (*squad.RunState, error) { return f.inner.LoadRun(wsDir) }
-
-func (f *faultingStore) SaveCheckpoint(wsDir string, cp squad.Checkpoint) error {
-	if f.failCheckpoint {
-		return fmt.Errorf("injected sqlite checkpoint save failure (superstep %d)", cp.Superstep)
-	}
-	return f.inner.SaveCheckpoint(wsDir, cp)
-}
-
-func (f *faultingStore) LoadLatestCheckpoint(wsDir string) (*squad.Checkpoint, error) {
-	return f.inner.LoadLatestCheckpoint(wsDir)
-}
-
-func (f *faultingStore) SaveGate(wsDir string, gate *squad.HitlGate) error {
-	return f.inner.SaveGate(wsDir, gate)
-}
-
-func (f *faultingStore) LoadGate(wsDir string) (*squad.HitlGate, error) {
-	return f.inner.LoadGate(wsDir)
-}
-
-func (f *faultingStore) SaveLeader(wsDir string, data []byte) error {
-	return f.inner.SaveLeader(wsDir, data)
-}
-
-func (f *faultingStore) LoadLeader(wsDir string) ([]byte, error) { return f.inner.LoadLeader(wsDir) }
-
-// eventTally 收集事件类型计数（断言没有成功事件）。
-type eventTally struct {
-	mu   sync.Mutex
-	seen []string
-	sink squad.ProgressSink
-}
-
-func newEventTally() *eventTally {
-	t := &eventTally{}
-	t.sink = squad.ProgressFunc(func(ev squad.Event) {
-		t.mu.Lock()
-		defer t.mu.Unlock()
-		t.seen = append(t.seen, ev.Type)
-	})
-	return t
-}
-
-func (t *eventTally) has(typ string) bool {
-	t.mu.Lock()
-	defer t.mu.Unlock()
-	for _, got := range t.seen {
-		if got == typ {
-			return true
-		}
-	}
-	return false
-}
-
-func (t *eventTally) types() []string {
-	t.mu.Lock()
-	defer t.mu.Unlock()
-	return append([]string(nil), t.seen...)
-}
-
-// TestSQLiteStoreSurfacesSaveFailures：SQLite 后端上的落盘故障必须可见——
-// 三档 checkpoint 模式下都不发 run_completed，运行要么报错、要么以失败收场。
-func TestSQLiteStoreSurfacesSaveFailures(t *testing.T) {
-	for _, mode := range []string{"sync", "async", "on_exit"} {
-		for _, fault := range []string{"checkpoint", "terminal"} {
-			t.Run(mode+"/"+fault, func(t *testing.T) {
-				runStore, err := OpenSQLiteStore(filepath.Join(t.TempDir(), "runs.db"))
-				if err != nil {
-					t.Fatal(err)
-				}
-				defer runStore.Close()
-				dir := t.TempDir()
-				reg := squad.NewRoleRegistry()
-				for _, role := range []string{"researcher", "writer", "reviewer"} {
-					if err := reg.Register(squad.RoleSpec{Role: role, Backend: "script"}); err != nil {
-						t.Fatal(err)
-					}
-				}
-				backends := squad.NewBackendResolver()
-				backends.RegisterInstance("script", stubExecutor{result: &squad.ExecResult{
-					Success: true, Logs: "ok",
-					Patches: map[string][]byte{"output.md": []byte(goodReportLite), "notes.md": []byte(goodReportLite), "fields.json": []byte("{}")},
-				}})
-				blueprints := squad.NewBlueprintRegistry()
-				if err := blueprints.Register(squad.Blueprint{
-					Key: "demo", Title: "故障演示", Match: []string{"故障演示"},
-					Steps: []squad.StepSpec{
-						{ID: "s1", Role: "researcher", Title: "调研", Targets: []string{"notes.md", "fields.json"}},
-						{ID: "s2", Role: "writer", Title: "成文", Targets: []string{"output.md"}},
-						{ID: "s3", Role: "reviewer", Title: "终审", Targets: []string{"output.md"}},
-					},
-				}); err != nil {
-					t.Fatal(err)
-				}
-				store := &faultingStore{inner: runStore,
-					failCheckpoint: fault == "checkpoint", failTerminal: fault == "terminal"}
-				tally := newEventTally()
-				eng, err := squad.NewEngine(squad.Options{
-					DataDir: dir, Registry: reg, Blueprints: blueprints,
-					Planner: squad.NewRulePlanner(), Backends: backends,
-					Sink: tally.sink, Store: store,
-				})
-				if err != nil {
-					t.Fatal(err)
-				}
-				res, err := eng.Run(context.Background(),
-					squad.Goal{Text: "做一个故障演示", HitlMode: squad.HitlAuto},
-					squad.RunOptions{CheckpointMode: mode})
-				if tally.has(squad.EvRunCompleted) {
-					t.Fatalf("run_completed emitted despite %s save failure: %v", fault, tally.types())
-				}
-				if fault == "checkpoint" && mode == "sync" {
-					if err == nil {
-						t.Fatalf("sync boundary failure must surface as error: %+v", res)
-					}
-					if !strings.Contains(err.Error(), "persistence failure") {
-						t.Fatalf("error not marked as persistence failure: %v", err)
-					}
-					return
-				}
-				if err != nil {
-					t.Fatalf("%s/%s: unexpected infra error: %v", mode, fault, err)
-				}
-				if res.Status != squad.RunFailed {
-					t.Fatalf("%s/%s: want failed, got %s", mode, fault, res.Status)
-				}
-				if !strings.Contains(res.Error, "persistence failure") {
-					t.Fatalf("%s/%s: failure not explained: %q", mode, fault, res.Error)
-				}
-				if !tally.has(squad.EvRunFailed) {
-					t.Fatalf("%s/%s: no run_failed event: %v", mode, fault, tally.types())
-				}
-			})
-		}
-	}
-}

squad/attempt_test.go

@@ -2,18 +2,12 @@ package squad
 
 import (
 	"context"
-	"encoding/json"
 	"os"
 	"path/filepath"
 	"sync"
 	"testing"
-	"time"
 )
 
-// jsonForTest 包装 json.Marshal 让测试里构型参数不拉锁。是仅为占用 import 走
-// 公共路径，不污染生产代码。
-func jsonForTest(v any) ([]byte, error) { return json.Marshal(v) }
-
 // perStepRecordingBackend 按子工作区归因计数（steps 的 role→tasks/NN_role 布局）。
 type perStepRecordingBackend struct {
 	mu    sync.Mutex
@@ -65,9 +59,6 @@ func newAttemptTestEngine(t *testing.T, backend *perStepRecordingBackend, dir st
 	return eng
 }
 
-// flipToInterrupted 把该步骤手动置为 running，同时清除 trusted terminal
-// record（模拟「崩溃于执行中」状态）。清除让该路径能够测试 F05 走的
-// 「无 record」分支。
 func flipToInterrupted(t *testing.T, workDir string, stepIdx int) {
 	t.Helper()
 	st, err := loadRunState(workDir)
@@ -84,16 +75,6 @@ func flipToInterrupted(t *testing.T, workDir string, stepIdx int) {
 	if err := st.save(); err != nil {
 		t.Fatal(err)
 	}
-	// 同时清除该步骤的 trusted terminal record（模拟未达记录写入点崩溃）。
-	terminals := loadStepTerminalsForWorkDir(workDir)
-	if terminals != nil {
-		stepID := st.Plan.Workflow.Steps[stepIdx].ID
-		if _, ok := terminals[stepID]; ok {
-			delete(terminals, stepID)
-			data, _ := jsonForTest(terminals)
-			_ = os.WriteFile(filepath.Join(workDir, "step_terminals.json"), data, 0o644)
-		}
-	}
 }
 
 // TestAttemptIDFormat：attempt id 稳定落盘且形如 runID/stepID/N（Q1 WP3）。
@@ -117,15 +98,8 @@ func TestAttemptIDFormat(t *testing.T) {
 }
 
 // TestKilledAtSuperstepBoundarySideEffectsOnce：杀进程于 superstep 边界后恢复，
-// 已完成步骤不重新执行（执行次数=1）。
-//
-// F05/U3b：running 崩溃、文件齐全、却无 trusted terminal record（U3b 前被视为
-// “副作用真发生”自动完成；U3b 后仅记录 step.Status=needs_reconciliation，
-// 后台不再自动 rerun/完成。本测试验证该迁变后的语义不变量：
-//
-//   - running 且无 record：需要 reconciliation，不隐默 rerun；
-//   - file presence 不替 trusted record 完成步骤。
-//   - 若事先写入 trusted record（手工合成），即便文件缺失也恢复为 completed。
+// 已完成步骤不重新执行（执行次数=1）；running 且产物齐全的步骤按 attempt 判定
+// 完成、不 rerun——Q1 验收第一条（WP3）。
 func TestKilledAtSuperstepBoundarySideEffectsOnce(t *testing.T) {
 	backend := &perStepRecordingBackend{}
 	dir := t.TempDir()
@@ -151,78 +125,22 @@ func TestKilledAtSuperstepBoundarySideEffectsOnce(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	// F05：需 reconciliation（auto 模式），不等于 completed。
-	if res.Status != RunWaitingReconciliation {
-		t.Fatalf("want waiting_reconciliation (auto mode), got %s", res.Status)
+	if res.Status != RunCompleted {
+		t.Fatalf("resume: %+v", res)
 	}
-	if res.Plan == nil || len(res.Plan.Steps) == 0 {
-		t.Fatal("missing plan")
-	}
-	var recon int
 	for _, s := range res.Plan.Steps {
-		if s.ID == "s2" && s.Status == StepStatusNeedsReconciliation {
-			recon++
+		if s.Status != "completed" && s.Status != "skipped" {
+			t.Fatalf("step %s not finished: %s", s.ID, s.Status)
 		}
-	}
-	if recon != 1 {
-		t.Fatalf("step s2 not in needs_reconciliation (got plan=%+v)", res.Plan.Steps)
-	}
-	// 重要：重复执行的次数不应发生隐默 rerun（本测试用 auto 模式运行走到 reconciliation）。
-	for _, s := range res.Plan.Steps {
 		got := backend.countOf(filepath.Base(filepath.FromSlash(s.SubWorkspace)))
 		if got != before[s.ID] {
-			t.Fatalf("step %s executed %d times after kill (before=%d)——running 步骤被隐默 rerun", s.ID, got, before[s.ID])
+			t.Fatalf("step %s executed %d times after kill (before=%d)——已完成/running 步骤被重复执行", s.ID, got, before[s.ID])
 		}
 	}
 }
 
-// TestTrustedRecordCompletesMissingFiles：trusted terminal record 存在 + 文件缺失
-// → 恢复为 completed（files 仍作诊断，state 是权威）。
-func TestTrustedRecordCompletesMissingFiles(t *testing.T) {
-	backend := &perStepRecordingBackend{}
-	dir := t.TempDir()
-	eng := newAttemptTestEngine(t, backend, dir)
-	res, err := eng.Run(context.Background(), Goal{Text: "trusted record 完成", HitlMode: HitlAuto}, RunOptions{})
-	if err != nil || res.Status != RunCompleted {
-		t.Fatalf("setup run: %v %+v", err, res)
-	}
-	// 制造「第二步走 running 但 trusted record 已存在」的场景：先写入 record，再
-	// flip 到 running + 删 file。
-	st, err := loadRunState(res.WorkDir)
-	if err != nil {
-		t.Fatal(err)
-	}
-	sub2 := filepath.FromSlash(st.Plan.Workflow.Steps[1].SubWorkspace)
-	flipToInterrupted(t, res.WorkDir, 1)
-	if err := os.Remove(filepath.Join(res.WorkDir, sub2, "output.md")); err != nil {
-		t.Fatal(err)
-	}
-	// 手工写入 trusted record（这能覆盖「后端状态已落盘但文件被外部删」该返回
-	// 已完成的合法语义）。
-	terminals := map[string]stepTerminalRecord{
-		"s2": {StepID: "s2", AttemptID: st.Plan.Workflow.Steps[1].AttemptID, Terminal: "completed", SideEffects: string(ExecSideEffectsNone), At: time.Now()},
-	}
-	data, _ := jsonForTest(terminals)
-	if err := os.WriteFile(filepath.Join(res.WorkDir, "step_terminals.json"), data, 0o644); err != nil {
-		t.Fatal(err)
-	}
-	before := backend.countOf(filepath.Base(sub2))
-	eng2 := newAttemptTestEngine(t, backend, dir)
-	res, err = eng2.Resume(context.Background(), res.RunID)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if res.Status != RunCompleted {
-		t.Fatalf("want completed via trusted record, got %s", res.Status)
-	}
-	if got := backend.countOf(filepath.Base(sub2)); got != before {
-		t.Fatalf("trusted-record step reran (before=%d after=%d)——record 需被信任为 completed", before, got)
-	}
-}
-
-// TestInterruptedMissingProductsEntersReconciliation：F05/U3b：running 崩溃 + 无
-// trusted record + 文件缺失 → auto 模式不再隐默 rerun，进入 waiting_reconciliation。
-func TestInterruptedMissingProductsEntersReconciliation(t *testing.T) {
+// TestInterruptedMissingProductsRetries：running 崩溃且产物缺失时按 auto 一次重试（WP3）。
+func TestInterruptedMissingProductsRetries(t *testing.T) {
 	backend := &perStepRecordingBackend{}
 	dir := t.TempDir()
 	eng := newAttemptTestEngine(t, backend, dir)
@@ -230,6 +148,7 @@ func TestInterruptedMissingProductsEntersReconciliation(t *testing.T) {
 	if err != nil || res.Status != RunCompleted {
 		t.Fatalf("setup run: %v %+v", err, res)
 	}
+	// 第二步停 running 并删掉其交付物（副作用未发生的形态）。
 	flipToInterrupted(t, res.WorkDir, 1)
 	st, err := loadRunState(res.WorkDir)
 	if err != nil {
@@ -239,24 +158,16 @@ func TestInterruptedMissingProductsEntersReconciliation(t *testing.T) {
 	if err := os.Remove(filepath.Join(res.WorkDir, sub, "output.md")); err != nil {
 		t.Fatal(err)
 	}
-	before := backend.countOf(filepath.Base(sub))
 	eng2 := newAttemptTestEngine(t, backend, dir)
 	res, err = eng2.Resume(context.Background(), res.RunID)
 	if err != nil {
 		t.Fatal(err)
 	}
-	if res.Status != RunWaitingReconciliation {
-		t.Fatalf("want waiting_reconciliation, got %s", res.Status)
+	if res.Status != RunCompleted {
+		t.Fatalf("resume: %+v", res)
 	}
-	// 后续 Resume 仍旧是 reconciliation，不再 rerun。
-	res2, err := eng2.Resume(context.Background(), res.RunID)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if res2.Status != RunWaitingReconciliation {
-		t.Fatalf("second resume shouldn't rerun, got %s", res2.Status)
-	}
-	if got := backend.countOf(filepath.Base(sub)); got != before {
-		t.Fatalf("missing-files step reran (before=%d after=%d)——record 缺失不应 rerun", before, got)
+	got := backend.countOf(filepath.Base(sub))
+	if got != 2 {
+		t.Fatalf("missing-product step executed %d times, want 2（auto 一次重试）", got)
 	}
 }

squad/backend.go

@@ -1,9 +1,7 @@
 package squad
 
 import (
-	"context"
 	"fmt"
-	"strings"
 	"sync"
 
 	"error-code-system/errcode"
@@ -18,20 +16,11 @@ type BackendFactory func(role RoleSpec) (Executor, error)
 // agent-native 自研内核，产品也可注册 "pi-agent" 等外部智能体工厂。
 // 未知名字 fail-closed，绝不静默落 stub。角色声明 MaxConcurrentTasks 时
 // 以并发限流包装实现对象级串行（每个角色一份缓存实例，限流跨步骤共享）。
-//
-// 缓存键含 ConfigIdentity 与工厂世代，故 BashPermission/工具配置收紧或工厂
-// 替换在下一次 Resolve 即生效。并发 gate 按「后端名+角色」独立于执行器版本
-// 持有：配置换代不会另起限流器而绕过角色上限。
-//
-// 边界：已在飞的任务不会被回滚或取消；收紧权限/降低并发只影响后续 Resolve
-// 与后续 Acquire。需要即时撤权的宿主应在每次工具调用前再校验。
 type BackendResolver struct {
 	mu         sync.RWMutex
 	factories  map[string]BackendFactory
 	underlying map[string]Executor // 无工厂直挂的后端实例
-	wrapped    map[string]Executor // 执行器配置身份 → 已包装实例缓存
-	factoryGen map[string]uint64   // Register/RegisterInstance 世代，替换即递增
-	gates      map[string]*roleConcurrencyGate
+	wrapped    map[string]Executor // 角色（含模型绑定）→ 已包装实例缓存
 }
 
 // NewBackendResolver 创建解析器并注册内置后端。
@@ -40,40 +29,23 @@ func NewBackendResolver() *BackendResolver {
 		factories:  make(map[string]BackendFactory),
 		underlying: make(map[string]Executor),
 		wrapped:    make(map[string]Executor),
-		factoryGen: make(map[string]uint64),
-		gates:      make(map[string]*roleConcurrencyGate),
 	}
 	r.RegisterInstance("stub", &StubExecutor{})
 	return r
 }
 
 // Register 注册后端工厂（名字如 "agent-native"、"pi-agent"）。
-// 替换同名工厂会使该后端下已缓存的包装执行器失效。
 func (r *BackendResolver) Register(name string, f BackendFactory) {
 	r.mu.Lock()
 	defer r.mu.Unlock()
 	r.factories[name] = f
-	r.invalidateLocked(name)
 }
 
 // RegisterInstance 直接注册后端实例（忽略角色级绑定）。
-// 替换同名实例会使该后端下已缓存的包装执行器失效；"stub" 语义保持不变。
 func (r *BackendResolver) RegisterInstance(name string, b Executor) {
 	r.mu.Lock()
 	defer r.mu.Unlock()
 	r.underlying[name] = b
-	r.invalidateLocked(name)
-}
-
-func (r *BackendResolver) invalidateLocked(name string) {
-	r.factoryGen[name]++
-	prefix := name + "\x00"
-	for k := range r.wrapped {
-		if strings.HasPrefix(k, prefix) {
-			delete(r.wrapped, k)
-		}
-	}
-	// gates 按角色身份保留：换工厂/配置不得丢掉进行中的并发计数。
 }
 
 // ListBackends 返回可用后端名（排序稳定）。
@@ -101,8 +73,8 @@ func (r *BackendResolver) ListBackends() []string {
 }
 
 // Resolve 解析角色的执行后端：Role.Backend → Options.DefaultBackend → "stub"。
-// MaxConcurrentTasks >= 1 时以共享角色 gate 包装（跨执行器版本共用；
-// 创建段持写锁防并行首解析竞态——否则并发步骤会各拿一份执行器）。
+// MaxConcurrentTasks > 1 时以 agent-native 并发限流包装（per-role 缓存，跨步骤共享；
+// 创建段持写锁防并行首解析竞态——否则并发步骤会各拿一份限流器）。
 func (r *BackendResolver) Resolve(role RoleSpec, defaultBackend string) (Executor, string, error) {
 	name := role.Backend
 	if name == "" {
@@ -111,44 +83,30 @@ func (r *BackendResolver) Resolve(role RoleSpec, defaultBackend string) (Executo
 	if name == "" {
 		name = "stub"
 	}
+	cacheKey := name + "\x00" + role.Role + "\x00" + role.Model
 
 	// 后端名快照在锁外获取（Resolve 持锁期间再调 ListBackends 会自死锁）。
 	names := r.ListBackends()
-
 	r.mu.RLock()
-	gen := r.factoryGen[name]
-	cacheKey := resolveCacheKey(name, role, gen)
 	cached, ok := r.wrapped[cacheKey]
+	factory, hasFactory := r.factories[name]
+	under, hasUnder := r.underlying[name]
 	r.mu.RUnlock()
 	if ok {
-		if max := role.MaxConcurrentTasks; max >= 1 {
-			r.mu.Lock()
-			if g := r.gates[limiterKey(name, role.Role)]; g != nil {
-				g.setMax(max)
-			}
-			r.mu.Unlock()
-		}
 		return cached, name, nil
 	}
 
 	r.mu.Lock()
 	defer r.mu.Unlock()
-	gen = r.factoryGen[name]
-	cacheKey = resolveCacheKey(name, role, gen)
 	// 双检：等写锁期间可能已被并发解析填充。
 	if b, ok := r.wrapped[cacheKey]; ok {
-		if max := role.MaxConcurrentTasks; max >= 1 {
-			if g := r.gates[limiterKey(name, role.Role)]; g != nil {
-				g.setMax(max)
-			}
-		}
 		return b, name, nil
 	}
 
 	var b Executor
-	if under, hasUnder := r.underlying[name]; hasUnder {
+	if hasUnder {
 		b = under
-	} else if factory, hasFactory := r.factories[name]; hasFactory {
+	} else if hasFactory {
 		var err error
 		b, err = factory(role)
 		if err != nil {
@@ -159,114 +117,9 @@ func (r *BackendResolver) Resolve(role RoleSpec, defaultBackend string) (Executo
 	}
 
 	// MaxConcurrentTasks>=1 即包装限流（1 = 对象级串行；0 = 不限）。
-	// gate 按角色身份共享，与 ConfigIdentity / 工厂世代无关。
 	if max := role.MaxConcurrentTasks; max >= 1 {
-		g := r.gateLocked(name, role.Role, max)
-		b = &gatedExecutor{inner: b, gate: g}
+		b = WrapWithConcurrencyLimit(b, max)
 	}
 	r.wrapped[cacheKey] = b
 	return b, name, nil
 }
-
-func resolveCacheKey(name string, role RoleSpec, gen uint64) string {
-	return name + "\x00" + role.Role + "\x00" + ConfigIdentity(role) + "\x00" +
-		fmt.Sprintf("%d", gen) + "\x00" + limiterStamp(role.MaxConcurrentTasks)
-}
-
-func limiterKey(backend, role string) string {
-	return backend + "\x00" + role
-}
-
-func (r *BackendResolver) gateLocked(backend, role string, max int) *roleConcurrencyGate {
-	key := limiterKey(backend, role)
-	g, ok := r.gates[key]
-	if !ok {
-		g = newRoleConcurrencyGate(max)
-		r.gates[key] = g
-		return g
-	}
-	g.setMax(max)
-	return g
-}
-
-// roleConcurrencyGate 是按角色共享的并发上限。
-// 执行器配置换代仍复用同一 gate，故旧版本在飞任务计入同一上限。
-type roleConcurrencyGate struct {
-	mu     sync.Mutex
-	max    int
-	active int
-	notify chan struct{} // 关闭并替换以唤醒等待者
-}
-
-func newRoleConcurrencyGate(max int) *roleConcurrencyGate {
-	if max <= 0 {
-		max = 1
-	}
-	return &roleConcurrencyGate{max: max, notify: make(chan struct{})}
-}
-
-func (g *roleConcurrencyGate) setMax(max int) {
-	if max <= 0 {
-		max = 1
-	}
-	g.mu.Lock()
-	defer g.mu.Unlock()
-	g.max = max
-	g.signalLocked()
-}
-
-func (g *roleConcurrencyGate) signalLocked() {
-	close(g.notify)
-	g.notify = make(chan struct{})
-}
-
-func (g *roleConcurrencyGate) acquire(ctx context.Context) error {
-	for {
-		g.mu.Lock()
-		if g.active < g.max {
-			g.active++
-			g.mu.Unlock()
-			return nil
-		}
-		notify := g.notify
-		g.mu.Unlock()
-		select {
-		case <-ctx.Done():
-			return ctx.Err()
-		case <-notify:
-		}
-	}
-}
-
-func (g *roleConcurrencyGate) release() {
-	g.mu.Lock()
-	defer g.mu.Unlock()
-	if g.active > 0 {
-		g.active--
-	}
-	g.signalLocked()
-}
-
-// gatedExecutor 把共享角色 gate 套在具体执行器上。
-type gatedExecutor struct {
-	inner Executor
-	gate  *roleConcurrencyGate
-}
-
-func (e *gatedExecutor) Name() string {
-	if e.inner == nil {
-		return "gated"
-	}
-	return e.inner.Name()
-}
-
-func (e *gatedExecutor) Execute(ctx context.Context, req ExecRequest) (*ExecResult, error) {
-	if err := e.gate.acquire(ctx); err != nil {
-		return &ExecResult{
-			Success:     false,
-			ErrorReason: fmt.Sprintf("agent concurrency wait canceled: %v", err),
-		}, err
-	}
-	defer e.gate.release()
-	return e.inner.Execute(ctx, req)
-}

squad/backend_test.go

@@ -1,405 +0,0 @@
-package squad
-
-import (
-	"context"
-	"sync"
-	"sync/atomic"
-	"testing"
-	"time"
-)
-
-type labeledExec struct {
-	label string
-	perm  string
-	tools string
-	delay time.Duration
-	enter func()
-	leave func()
-}
-
-func (e *labeledExec) Name() string { return e.label }
-
-func (e *labeledExec) Execute(ctx context.Context, req ExecRequest) (*ExecResult, error) {
-	if e.enter != nil {
-		e.enter()
-	}
-	if e.leave != nil {
-		defer e.leave()
-	}
-	if e.delay > 0 {
-		select {
-		case <-ctx.Done():
-			return nil, ctx.Err()
-		case <-time.After(e.delay):
-		}
-	}
-	return &ExecResult{Success: true, Logs: e.label + "|" + e.perm + "|" + e.tools}, nil
-}
-
-func TestConfigIdentityChangesWithBashAndTools(t *testing.T) {
-	a := RoleSpec{Role: "r", Model: "m", BashPermission: BashPermissionAllow}
-	d := RoleSpec{Role: "r", Model: "m", BashPermission: BashPermissionDeny}
-	if ConfigIdentity(a) == ConfigIdentity(d) {
-		t.Fatal("allow and deny must differ")
-	}
-	t1 := RoleSpec{Role: "r", Model: "m", Tools: []string{"a", "b"}}
-	t2 := RoleSpec{Role: "r", Model: "m", Tools: []string{"b", "a"}}
-	if ConfigIdentity(t1) != ConfigIdentity(t2) {
-		t.Fatal("tool order must not change identity")
-	}
-	t3 := RoleSpec{Role: "r", Model: "m", Tools: []string{"a", "c"}}
-	if ConfigIdentity(t1) == ConfigIdentity(t3) {
-		t.Fatal("tool set change must change identity")
-	}
-	same := RoleSpec{Role: "r", Model: "m", BashPermission: BashPermissionAllow, MaxConcurrentTasks: 1}
-	bump := RoleSpec{Role: "r", Model: "m", BashPermission: BashPermissionAllow, MaxConcurrentTasks: 8}
-	if ConfigIdentity(same) != ConfigIdentity(bump) {
-		t.Fatal("MaxConcurrentTasks must not affect ConfigIdentity")
-	}
-}
-
-func TestResolveAllowDenyTighteningAndLoosening(t *testing.T) {
-	var builds atomic.Int32
-	r := NewBackendResolver()
-	r.Register("cap", func(role RoleSpec) (Executor, error) {
-		builds.Add(1)
-		return &labeledExec{label: "cap", perm: role.BashPermission}, nil
-	})
-
-	allow := RoleSpec{Role: "writer", Backend: "cap", Model: "m1", BashPermission: BashPermissionAllow}
-	e1, _, err := r.Resolve(allow, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, _ := e1.Execute(context.Background(), ExecRequest{})
-	if res.Logs != "cap|allow|" {
-		t.Fatalf("want allow executor, got %q", res.Logs)
-	}
-
-	deny := RoleSpec{Role: "writer", Backend: "cap", Model: "m1", BashPermission: BashPermissionDeny}
-	e2, _, err := r.Resolve(deny, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, _ = e2.Execute(context.Background(), ExecRequest{})
-	if res.Logs != "cap|deny|" {
-		t.Fatalf("allow→deny must take effect on next Resolve, got %q", res.Logs)
-	}
-	if e1 == e2 {
-		t.Fatal("tightened config must not reuse cached allow executor")
-	}
-
-	e3, _, err := r.Resolve(allow, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, _ = e3.Execute(context.Background(), ExecRequest{})
-	if res.Logs != "cap|allow|" {
-		t.Fatalf("deny→allow must take effect, got %q", res.Logs)
-	}
-	if builds.Load() < 2 {
-		t.Fatalf("factory must rebuild on permission change, builds=%d", builds.Load())
-	}
-}
-
-func TestResolveFactoryReplacementTakesEffect(t *testing.T) {
-	r := NewBackendResolver()
-	r.Register("cap", func(role RoleSpec) (Executor, error) {
-		return &labeledExec{label: "v1", perm: role.BashPermission}, nil
-	})
-	role := RoleSpec{Role: "writer", Backend: "cap", Model: "m", BashPermission: BashPermissionAllow}
-	e1, _, err := r.Resolve(role, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, _ := e1.Execute(context.Background(), ExecRequest{})
-	if res.Logs != "v1|allow|" {
-		t.Fatalf("got %q", res.Logs)
-	}
-
-	r.Register("cap", func(role RoleSpec) (Executor, error) {
-		return &labeledExec{label: "v2", perm: role.BashPermission}, nil
-	})
-	e2, _, err := r.Resolve(role, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, _ = e2.Execute(context.Background(), ExecRequest{})
-	if res.Logs != "v2|allow|" {
-		t.Fatalf("factory replacement must take effect, got %q", res.Logs)
-	}
-}
-
-func TestResolveToolConfigChangeTakesEffect(t *testing.T) {
-	r := NewBackendResolver()
-	r.Register("cap", func(role RoleSpec) (Executor, error) {
-		tools := ""
-		if len(role.Tools) > 0 {
-			tools = role.Tools[0]
-		}
-		return &labeledExec{label: "cap", tools: tools}, nil
-	})
-	base := RoleSpec{Role: "writer", Backend: "cap", Model: "m", Tools: []string{"read"}}
-	e1, _, err := r.Resolve(base, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, _ := e1.Execute(context.Background(), ExecRequest{})
-	if res.Logs != "cap||read" {
-		t.Fatalf("got %q", res.Logs)
-	}
-	next := RoleSpec{Role: "writer", Backend: "cap", Model: "m", Tools: []string{"write"}}
-	e2, _, err := r.Resolve(next, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, _ = e2.Execute(context.Background(), ExecRequest{})
-	if res.Logs != "cap||write" {
-		t.Fatalf("tool-config change must take effect, got %q", res.Logs)
-	}
-}
-
-func TestResolveUnchangedConfigReusesCache(t *testing.T) {
-	var builds atomic.Int32
-	r := NewBackendResolver()
-	r.Register("cap", func(role RoleSpec) (Executor, error) {
-		builds.Add(1)
-		return &labeledExec{label: "cap", perm: role.BashPermission}, nil
-	})
-	role := RoleSpec{Role: "writer", Backend: "cap", Model: "m", BashPermission: BashPermissionAsk}
-	e1, _, err := r.Resolve(role, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	e2, _, err := r.Resolve(role, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	if e1 != e2 {
-		t.Fatal("unchanged config must reuse cached executor")
-	}
-	if builds.Load() != 1 {
-		t.Fatalf("want 1 factory call, got %d", builds.Load())
-	}
-}
-
-func TestResolveParallelFirstDoesNotDuplicate(t *testing.T) {
-	var builds atomic.Int32
-	r := NewBackendResolver()
-	r.Register("cap", func(role RoleSpec) (Executor, error) {
-		builds.Add(1)
-		time.Sleep(20 * time.Millisecond) // widen race window
-		return &labeledExec{label: "cap"}, nil
-	})
-	role := RoleSpec{Role: "writer", Backend: "cap", Model: "m", MaxConcurrentTasks: 2}
-
-	const n = 32
-	var wg sync.WaitGroup
-	got := make([]Executor, n)
-	errs := make([]error, n)
-	wg.Add(n)
-	for i := 0; i < n; i++ {
-		i := i
-		go func() {
-			defer wg.Done()
-			got[i], _, errs[i] = r.Resolve(role, "")
-		}()
-	}
-	wg.Wait()
-	for i, err := range errs {
-		if err != nil {
-			t.Fatalf("resolve %d: %v", i, err)
-		}
-	}
-	for i := 1; i < n; i++ {
-		if got[i] != got[0] {
-			t.Fatalf("parallel first Resolve must share one executor")
-		}
-	}
-	if builds.Load() != 1 {
-		t.Fatalf("want single factory construction, got %d", builds.Load())
-	}
-}
-
-func TestSharedGateAcrossVersionsAndLimiterChange(t *testing.T) {
-	var (
-		mu      sync.Mutex
-		current int
-		peak    int
-	)
-	hold := make(chan struct{})
-	entered := make(chan struct{}, 8)
-	trackEnter := func() {
-		mu.Lock()
-		current++
-		if current > peak {
-			peak = current
-		}
-		mu.Unlock()
-		entered <- struct{}{}
-		<-hold
-	}
-	trackLeave := func() {
-		mu.Lock()
-		current--
-		mu.Unlock()
-	}
-
-	r := NewBackendResolver()
-	r.Register("cap", func(role RoleSpec) (Executor, error) {
-		return &labeledExec{
-			label: "cap",
-			perm:  role.BashPermission,
-			enter: trackEnter,
-			leave: trackLeave,
-		}, nil
-	})
-
-	allow := RoleSpec{Role: "writer", Backend: "cap", Model: "m", BashPermission: BashPermissionAllow, MaxConcurrentTasks: 1}
-	eOld, _, err := r.Resolve(allow, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	errCh := make(chan error, 4)
-	go func() { _, err := eOld.Execute(context.Background(), ExecRequest{}); errCh <- err }()
-	<-entered // old holds the sole slot
-
-	deny := RoleSpec{Role: "writer", Backend: "cap", Model: "m", BashPermission: BashPermissionDeny, MaxConcurrentTasks: 1}
-	eNew, _, err := r.Resolve(deny, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	if eOld == eNew {
-		t.Fatal("version bump must produce a new executor")
-	}
-
-	secondDone := make(chan struct{})
-	go func() {
-		_, err := eNew.Execute(context.Background(), ExecRequest{})
-		errCh <- err
-		close(secondDone)
-	}()
-	select {
-	case <-secondDone:
-		t.Fatal("new-version execute must wait on shared role cap while old is in-flight")
-	case <-time.After(40 * time.Millisecond):
-	}
-
-	// Raising MaxConcurrentTasks must update the shared gate (not spawn a bypassing limiter).
-	deny2 := RoleSpec{Role: "writer", Backend: "cap", Model: "m", BashPermission: BashPermissionDeny, MaxConcurrentTasks: 2}
-	eLim, _, err := r.Resolve(deny2, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	go func() { _, err := eLim.Execute(context.Background(), ExecRequest{}); errCh <- err }()
-
-	// After setMax(2): waiting eNew can enter (active=2). eLim must still wait.
-	select {
-	case <-entered: // eNew
-	case <-time.After(time.Second):
-		t.Fatal("limiter raise must admit the waiting new-version task")
-	}
-	select {
-	case <-entered:
-		t.Fatal("third task must not bypass shared cap after limiter change")
-	case <-time.After(40 * time.Millisecond):
-	}
-
-	close(hold)
-	for i := 0; i < 3; i++ {
-		if err := <-errCh; err != nil {
-			t.Fatal(err)
-		}
-	}
-	mu.Lock()
-	p := peak
-	mu.Unlock()
-	if p > 2 {
-		t.Fatalf("shared gate must cap total in-flight across versions; peak=%d", p)
-	}
-}
-
-func TestInFlightOldVersionCannotExceedCap(t *testing.T) {
-	var (
-		mu      sync.Mutex
-		current int
-		peak    int
-	)
-	enter := func() {
-		mu.Lock()
-		current++
-		if current > peak {
-			peak = current
-		}
-		mu.Unlock()
-	}
-	leave := func() {
-		mu.Lock()
-		current--
-		mu.Unlock()
-	}
-
-	r := NewBackendResolver()
-	r.Register("cap", func(role RoleSpec) (Executor, error) {
-		return &labeledExec{
-			label: "cap",
-			perm:  role.BashPermission,
-			delay: 80 * time.Millisecond,
-			enter: enter,
-			leave: leave,
-		}, nil
-	})
-
-	v1 := RoleSpec{Role: "writer", Backend: "cap", Model: "m", BashPermission: BashPermissionAllow, MaxConcurrentTasks: 2}
-	e1, _, err := r.Resolve(v1, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	v2 := RoleSpec{Role: "writer", Backend: "cap", Model: "m", BashPermission: BashPermissionDeny, MaxConcurrentTasks: 2}
-	e2, _, err := r.Resolve(v2, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-
-	var wg sync.WaitGroup
-	// 2 from old version + 2 from new version; shared cap=2 ⇒ peak must stay ≤2.
-	for _, e := range []Executor{e1, e1, e2, e2} {
-		wg.Add(1)
-		e := e
-		go func() {
-			defer wg.Done()
-			_, _ = e.Execute(context.Background(), ExecRequest{})
-		}()
-	}
-	wg.Wait()
-
-	mu.Lock()
-	p := peak
-	mu.Unlock()
-	if p > 2 {
-		t.Fatalf("in-flight old+new must not exceed role cap; peak=%d", p)
-	}
-}
-
-func TestRegisterInstanceStubSemantics(t *testing.T) {
-	r := NewBackendResolver()
-	e, name, err := r.Resolve(RoleSpec{Role: "x"}, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	if name != "stub" {
-		t.Fatalf("want stub, got %s", name)
-	}
-	if e.Name() != "stub" {
-		t.Fatalf("want stub executor, got %s", e.Name())
-	}
-	custom := &labeledExec{label: "stub-custom"}
-	r.RegisterInstance("stub", custom)
-	e2, _, err := r.Resolve(RoleSpec{Role: "x"}, "")
-	if err != nil {
-		t.Fatal(err)
-	}
-	if e2.Name() != "stub-custom" {
-		t.Fatalf("RegisterInstance stub replace must take effect, got %s", e2.Name())
-	}
-}

squad/checkpoint_test.go

@@ -2,11 +2,9 @@ package squad
 
 import (
 	"context"
-	"fmt"
 	"os"
 	"path/filepath"
 	"strings"
-	"sync"
 	"testing"
 )
 
@@ -113,225 +111,3 @@ func TestGraphContinuityRejectsStructuralChange(t *testing.T) {
 		t.Fatalf("expected continuity rejection, got %v", err)
 	}
 }
-
-// ---- F11c：落盘失败的诚实性 ----
-
-// faultStore 包装底层 Store 注入落盘故障，其余路径照常透传。
-// failCheckpoint：任何 SaveCheckpoint 失败（async 档在出口 flush 时暴露）。
-// failTerminalSave：Meta 进入终态时的 SaveRun 失败（finalize/manualPass 的收口写）。
-type faultStore struct {
-	Store
-	mu               sync.Mutex
-	failCheckpoint   bool
-	failTerminalSave bool
-	checkpointWrites int
-	terminalAttempts int
-}
-
-func (f *faultStore) SaveCheckpoint(wsDir string, cp Checkpoint) error {
-	f.mu.Lock()
-	f.checkpointWrites++
-	shouldFail := f.failCheckpoint
-	f.mu.Unlock()
-	if shouldFail {
-		return fmt.Errorf("injected checkpoint write failure (superstep %d)", cp.Superstep)
-	}
-	return f.Store.SaveCheckpoint(wsDir, cp)
-}
-
-func (f *faultStore) SaveRun(st *RunState) error {
-	f.mu.Lock()
-	f.terminalAttempts++
-	shouldFail := f.failTerminalSave && RunStatus(st.Meta.Status).Terminal()
-	f.mu.Unlock()
-	if shouldFail {
-		return fmt.Errorf("injected terminal run state failure (%s)", st.Meta.Status)
-	}
-	return f.Store.SaveRun(st)
-}
-
-func (f *faultStore) writes() int {
-	f.mu.Lock()
-	defer f.mu.Unlock()
-	return f.checkpointWrites
-}
-
-// eventRecorder 按序收集引擎事件（顺序断言用）。
-type eventRecorder struct {
-	mu   sync.Mutex
-	evs  []Event
-	sink ProgressSink
-}
-
-func newEventRecorder() *eventRecorder {
-	r := &eventRecorder{}
-	r.sink = ProgressFunc(func(ev Event) {
-		r.mu.Lock()
-		defer r.mu.Unlock()
-		r.evs = append(r.evs, ev)
-	})
-	return r
-}
-
-func (r *eventRecorder) types() []string {
-	r.mu.Lock()
-	defer r.mu.Unlock()
-	out := make([]string, 0, len(r.evs))
-	for _, ev := range r.evs {
-		out = append(out, ev.Type)
-	}
-	return out
-}
-
-func (r *eventRecorder) has(t string) bool {
-	for _, got := range r.types() {
-		if got == t {
-			return true
-		}
-	}
-	return false
-}
-
-// newTestEngineWithStore 与 newTestEngine 同装配，额外注入 Store 与事件收集 sink。
-func newTestEngineWithStore(t *testing.T, store Store, rec *eventRecorder) *SquadEngine {
-	t.Helper()
-	dir := t.TempDir()
-	reg := NewRoleRegistry()
-	for _, role := range []string{"researcher", "writer", "reviewer"} {
-		if err := reg.Register(RoleSpec{Role: role, Backend: "script"}); err != nil {
-			t.Fatal(err)
-		}
-	}
-	backends := NewBackendResolver()
-	backends.RegisterInstance("script", &scriptBackend{script: map[string]map[string]string{
-		"01_researcher": {"notes.md": goodReport, "fields.json": "{}"},
-		"02_writer":     {"output.md": goodReport},
-		"03_reviewer":   {"output.md": goodReport},
-	}})
-	eng, err := NewEngine(Options{DataDir: dir, Registry: reg,
-		Planner: mockScriptPlanner(baseSteps()...), Backends: backends,
-		Sink: rec.sink, Store: store})
-	if err != nil {
-		t.Fatal(err)
-	}
-	return eng
-}
-
-// assertNoCleanCompletion：落盘失败后不得再以「干净完成」面世——
-// 没有 run_completed 事件、也没有 completed/degraded 的返回值。
-func assertNoCleanCompletion(t *testing.T, res *RunResult, err error, rec *eventRecorder) {
-	t.Helper()
-	if rec.has(EvRunCompleted) {
-		t.Fatalf("run_completed emitted after persistence failure: %v", rec.types())
-	}
-	if res != nil {
-		if res.Status == RunCompleted || res.Status == RunDegraded {
-			t.Fatalf("clean completion after persistence failure: status=%s", res.Status)
-		}
-		if res.Status != RunFailed {
-			t.Fatalf("status = %s, want failed or an error return", res.Status)
-		}
-		if !strings.Contains(res.Error, persistenceFailurePrefix) {
-			t.Fatalf("failure not explained in result: %q", res.Error)
-		}
-	}
-	if err != nil && !strings.Contains(err.Error(), persistenceFailurePrefix) {
-		t.Fatalf("error does not name the persistence failure: %v", err)
-	}
-	if res == nil && err == nil {
-		t.Fatal("persistence failure vanished: neither result nor error")
-	}
-}
-
-// TestCheckpointFlushFailureIsVisible：出口 flush 的落盘失败必须可见，
-// 且先于任何完成事件/返回（三档 sync/async/on_exit 都一样）。
-func TestCheckpointFlushFailureIsVisible(t *testing.T) {
-	for _, mode := range []string{"sync", "async", "on_exit"} {
-		t.Run(mode, func(t *testing.T) {
-			store := &faultStore{Store: FileStore{}, failCheckpoint: true}
-			rec := newEventRecorder()
-			eng := newTestEngineWithStore(t, store, rec)
-			res, err := eng.Run(context.Background(), Goal{Text: "快照"}, RunOptions{CheckpointMode: mode})
-			if store.writes() == 0 {
-				t.Fatal("test did not reach any checkpoint write")
-			}
-			assertNoCleanCompletion(t, res, err, rec)
-			// 失败在推进退出处如实标注；async/on_exit 由 finalizeEvent 发 run_failed，
-			// sync 在第一个边界就报错返回。
-			if mode == "sync" {
-				if err == nil {
-					t.Fatalf("sync boundary write failure must surface as error, got %+v", res)
-				}
-				return
-			}
-			if res == nil || res.Status != RunFailed {
-				t.Fatalf("%s: want failed result, got %+v err=%v", mode, res, err)
-			}
-			if !rec.has(EvRunFailed) {
-				t.Fatalf("%s: no run_failed event: %v", mode, rec.types())
-			}
-		})
-	}
-}
-
-// TestTerminalSaveFailureBlocksCompletion：终态运行态写盘失败时不得宣布完成
-// （事件与返回值都必须可区分）；业务动作已发生（交付物文件仍在），但不声称成功。
-func TestTerminalSaveFailureBlocksCompletion(t *testing.T) {
-	for _, mode := range []string{"sync", "async", "on_exit"} {
-		t.Run(mode, func(t *testing.T) {
-			store := &faultStore{Store: FileStore{}, failTerminalSave: true}
-			rec := newEventRecorder()
-			eng := newTestEngineWithStore(t, store, rec)
-			res, err := eng.Run(context.Background(), Goal{Text: "快照"}, RunOptions{CheckpointMode: mode})
-			if err != nil {
-				t.Fatalf("terminal save failure must not abort the run as infra error: %v", err)
-			}
-			assertNoCleanCompletion(t, res, err, rec)
-			if res.Status != RunFailed {
-				t.Fatalf("%s: want failed, got %s", mode, res.Status)
-			}
-			if !rec.has(EvRunFailed) {
-				t.Fatalf("%s: no run_failed event: %v", mode, rec.types())
-			}
-			// 完成事件必须晚于/缺席于失败：事件序里终态只有 run_failed。
-			types := rec.types()
-			lastTerminal := ""
-			for _, ty := range types {
-				switch ty {
-				case EvRunFailed, EvRunCompleted, EvRunCancelled:
-					lastTerminal = ty
-				}
-			}
-			if lastTerminal != EvRunFailed {
-				t.Fatalf("%s: last terminal event = %q, want run_failed", mode, lastTerminal)
-			}
-			// 业务动作已发生且未被回滚：交付物文件确实写出来了（但运行不声称完成）。
-			if _, err := os.Stat(filepath.Join(res.WorkDir, "output.md")); err != nil {
-				t.Fatalf("deliverable written before terminal save should remain: %v", err)
-			}
-		})
-	}
-}
-
-// TestCheckpointFlushSucceedsInAllModes（对照）：故障注入关闭时三档都正常收口，
-// 出口快照确实落盘（避免上面的失败断言因「压根没写」而空过）。
-func TestCheckpointFlushSucceedsInAllModes(t *testing.T) {
-	for _, mode := range []string{"sync", "async", "on_exit"} {
-		t.Run(mode, func(t *testing.T) {
-			store := &faultStore{Store: FileStore{}}
-			rec := newEventRecorder()
-			eng := newTestEngineWithStore(t, store, rec)
-			res, err := eng.Run(context.Background(), Goal{Text: "快照"}, RunOptions{CheckpointMode: mode})
-			if err != nil || res.Status != RunCompleted {
-				t.Fatalf("%s: %v %+v", mode, err, res)
-			}
-			if !rec.has(EvRunCompleted) {
-				t.Fatalf("%s: no run_completed: %v", mode, rec.types())
-			}
-			latest, err := FileStore{}.LoadLatestCheckpoint(res.WorkDir)
-			if err != nil || latest == nil || latest.Superstep < 3 {
-				t.Fatalf("%s: exit snapshot not flushed: %v %+v", mode, err, latest)
-			}
-		})
-	}
-}

squad/config_identity.go

@@ -1,45 +0,0 @@
-package squad
-
-import (
-	"crypto/sha256"
-	"encoding/hex"
-	"sort"
-	"strings"
-)
-
-// ConfigIdentity 返回角色中「塑造 Executor 装配」字段的稳定摘要。
-// 含 BashPermission / Model / ToolProfile / Tools / Skills；不含
-// MaxConcurrentTasks（并发上限由解析器按角色身份单独持有，与执行器版本解耦）。
-func ConfigIdentity(role RoleSpec) string {
-	var b strings.Builder
-	b.WriteString("bash=")
-	b.WriteString(role.BashPermission)
-	b.WriteString("\x00model=")
-	b.WriteString(role.Model)
-	b.WriteString("\x00profile=")
-	b.WriteString(role.ToolProfile)
-	b.WriteString("\x00tools=")
-	b.WriteString(stableJoin(role.Tools))
-	b.WriteString("\x00skills=")
-	b.WriteString(stableJoin(role.Skills))
-	sum := sha256.Sum256([]byte(b.String()))
-	return hex.EncodeToString(sum[:16])
-}
-
-func stableJoin(items []string) string {
-	if len(items) == 0 {
-		return ""
-	}
-	cp := append([]string(nil), items...)
-	sort.Strings(cp)
-	return strings.Join(cp, ",")
-}
-
-// limiterStamp 区分「启用限流」与「不限流」；具体上限数值不进执行器缓存键，
-// 以免调大/调小 MaxConcurrentTasks 时换掉共享 gate。
-func limiterStamp(max int) string {
-	if max >= 1 {
-		return "lim"
-	}
-	return "unlim"
-}

squad/engine.go

@@ -973,18 +973,9 @@ func (e *SquadEngine) manualPass(st *RunState, decision RespondDecision) (*RunRe
 	wf := st.Workflow()
 	wf.Status = string(status)
 	st.setWorkflow(wf)
-	// F11c：终态落盘先于完成事件——写不进去不得宣布完成（业务动作不回滚）。
-	if serr := st.save(); serr != nil {
-		msg := persistenceFailureText("save terminal run state", serr)
-		return &RunResult{RunID: st.Meta.ID, Status: RunFailed, Plan: planPtr(st),
-			WorkDir: st.Meta.WorkDir, Quality: report, Error: msg, Summary: msg}, nil
-	}
+	_ = st.save()
 	manifest := buildManifest(st, "output.md", *report)
-	if merr := writeJSON(filepath.Join(st.Meta.WorkDir, "manifest.json"), manifest); merr != nil {
-		msg := persistenceFailureText("write manifest", merr)
-		return &RunResult{RunID: st.Meta.ID, Status: RunFailed, Plan: planPtr(st),
-			WorkDir: st.Meta.WorkDir, Quality: report, Error: msg, Summary: msg}, nil
-	}
+	_ = writeJSON(filepath.Join(st.Meta.WorkDir, "manifest.json"), manifest)
 	e.deps.Sink.Emit(Event{Type: EvRunCompleted, RunID: st.Meta.ID, Text: string(status), At: time.Now()})
 	return &RunResult{RunID: st.Meta.ID, Status: status, Plan: planPtr(st),
 		Deliverable: "output.md", Quality: report, WorkDir: st.Meta.WorkDir,
@@ -998,19 +989,10 @@ func (e *SquadEngine) cancelRun(st *RunState, reason string) *RunResult {
 	wf := st.Workflow()
 	wf.Status = "cancelled"
 	st.setWorkflow(wf)
-	// 取消语义不变（运行确实被中止），但取消态落盘失败必须可见，不静默。
-	saveErr := st.save()
-	if saveErr != nil {
-		e.deps.Sink.Emit(Event{Type: EvPhase, RunID: st.Meta.ID,
-			Text: persistenceFailureText("save cancelled run state", saveErr), At: time.Now()})
-	}
+	_ = st.save()
 	e.deps.Sink.Emit(Event{Type: EvRunCancelled, RunID: st.Meta.ID, Text: reason, At: time.Now()})
-	out := &RunResult{RunID: st.Meta.ID, Status: RunCancelled, Plan: planPtr(st),
+	return &RunResult{RunID: st.Meta.ID, Status: RunCancelled, Plan: planPtr(st),
 		Summary: reason, WorkDir: st.Meta.WorkDir}
-	if saveErr != nil {
-		out.Error = persistenceFailureText("save cancelled run state", saveErr)
-	}
-	return out
 }
 
 // failStepResume 权限拒绝后的失败步骤处理（关键失败/非关键降级）。
@@ -1069,129 +1051,73 @@ func markStep(st *RunState, id, status, errText string) bool {
 }
 
 // recoverInterrupted 处理崩溃时停在 running 的步骤。
-// interactive 挂起确认/对账，不自动再执行；auto 允许一次重试；disabled 直接失败。
-//
-// F05/U3b：恢复是否判定 completed 不再单纯依据「文件齐全」——trusted terminal
-// record 是单一权威：仅当 step_terminals.json 中存在该步骤且 Terminal=completed
-// 时才能恢复完成（即便文件缺失也照恢复，文件用于诊断）；其余情形按 record +
-// 文件状态联合判定：
-//
-//   - 有 trusted record（completed）+ 缺失文件 -> completed（files 诊断信息）
-//   - 有 trusted record + 现存文件 -> completed（一致）
-//   - 有 trusted record + SideEffects=none + 缺失文件 -> 一次重试 pending
-//   - 无 record + 文件齐全 -> needs_reconciliation（不自动 rerun，不假装完成）
-//   - 无 record + 文件缺失 -> needs_reconciliation
-//   - record 表明 Canceled/Aborted -> 走 cancel 路径
-//
-// retention 期间产物缺失若 record SideEffects=possible，禁止无条件 rerun ——
-// 必须经人工 reconciliation 才能继续。
+// interactive 挂起确认，不自动再执行；auto 允许一次重试；disabled 直接失败。
 func (e *SquadEngine) recoverInterrupted(ctx context.Context, st *RunState) (*RunResult, error) {
-	terminals := loadStepTerminalsForWorkDir(st.Meta.WorkDir)
-	drv := &flowDriver{opts: e.deps}
 	for i := range st.Plan.Workflow.Steps {
 		step := st.Plan.Workflow.Steps[i]
 		if step.Status != "running" {
 			continue
 		}
-		subDir := filepath.Join(st.Meta.WorkDir, filepath.FromSlash(orDefault(step.SubWorkspace, fmt.Sprintf("tasks/%02d_%s", i+1, step.Role))))
-		targets := step.Targets
-		if len(targets) == 0 {
-			targets = []string{"output.md"}
-		}
-		missing := missingTargets(subDir, targets)
-		record, hasRecord := terminals[step.ID]
-
-		// 1. trusted terminal record 存在 -> 单一权威。
-		if hasRecord && record.Terminal == "completed" && !record.Canceled && !record.Aborted {
-			step.Status = "completed"
-			step.Error = ""
-			st.Plan.Workflow.Steps[i] = step
-			st.mu.Lock()
-			_ = mergeStepArtifacts(&st.BB, subDir)
-			err := st.save()
-			st.mu.Unlock()
-			if err != nil {
-				return nil, err
-			}
-			drv.emit(st, Event{Type: EvStepCompleted, StepID: step.ID, Role: step.Role,
-				Text: fmt.Sprintf("中断恢复：trusted terminal record（attempt=%s, side_effects=%s）判定完成",
-					record.AttemptID, orDefault(record.SideEffects, "possible"))})
-			continue
-		}
-		// 1b. record 表明已取消/中止 -> cancel 路径（不等同业务失败）。
-		if hasRecord && (record.Canceled || record.Aborted) {
-			step.Status = "failed"
-			if record.Canceled {
-				step.Error = "interrupted: canceled"
-			} else {
-				step.Error = "interrupted: aborted"
-			}
-			st.Plan.Workflow.Steps[i] = step
-			if err := st.save(); err != nil {
-				return nil, err
-			}
-			out, _, ferr := drv.failStep(ctx, st, i)
-			return out, ferr
-		}
-		// 1c. record 表明 SideEffects=none + 缺失文件 -> 仅一次重试 pending
-		// （旧 WP3 路径保留以兼容纯读取类步骤）。
-		if hasRecord && record.SideEffects == string(ExecSideEffectsNone) && len(missing) > 0 && step.Attempts < 2 {
-			step.Attempts++
-			step.Status = "pending"
-			step.Error = ""
-			// F05：复用上一次 attempt 的 ExecutionID（已在 step.Terminal 落 record）。
-			if record.AttemptID != "" {
-				step.AttemptID = record.AttemptID
-			}
-			st.Plan.Workflow.Steps[i] = step
-			if err := st.save(); err != nil {
-				return nil, err
-			}
-			continue
-		}
-
-		// 2. 无 trusted record —— 旧路径不准直接翻 completed，必须进入 reconciliation。
-		// 区分 HITL 模式：interactive 仍合成 HITL hold（同 hitlConfirmation 通道），
-		// auto / disabled 走显式 needs_reconciliation（不自动 rerun 不假装成功）。
-		reason := "no trusted terminal record"
-		if len(missing) == 0 {
-			reason = "no trusted terminal record despite deliverables present"
-		} else {
-			reason = "no trusted terminal record and deliverables missing"
-		}
-		step.Status = StepStatusNeedsReconciliation
-		step.Error = reason
-		st.Plan.Workflow.Steps[i] = step
-		if err := st.save(); err != nil {
-			return nil, err
-		}
 		switch st.policy().mode() {
-		case HitlInteractive:
+		case HitlDisabled:
+			st.Plan.Workflow.Steps[i].Status = "failed"
+			st.Plan.Workflow.Steps[i].Error = "interrupted"
+			if err := st.save(); err != nil {
+				return nil, err
+			}
+			drv := &flowDriver{opts: e.deps}
+			out, _, err := drv.failStep(ctx, st, i)
+			return out, err
+		case HitlAuto:
+			if step.Attempts >= 2 {
+				st.Plan.Workflow.Steps[i].Status = "failed"
+				st.Plan.Workflow.Steps[i].Error = "interrupted twice"
+				if err := st.save(); err != nil {
+					return nil, err
+				}
+				drv := &flowDriver{opts: e.deps}
+				out, _, err := drv.failStep(ctx, st, i)
+				return out, err
+			}
+			// Q1 WP3 产物校验：步骤在 running 崩溃，但其交付物已全部落盘——
+			// 副作用实际发生过，直接判定完成而不是重新执行（副作用次数=1）。
+			subDir := filepath.Join(st.Meta.WorkDir, filepath.FromSlash(orDefault(step.SubWorkspace, fmt.Sprintf("tasks/%02d_%s", i+1, step.Role))))
+			targets := step.Targets
+			if len(targets) == 0 {
+				targets = []string{"output.md"}
+			}
+			if missingTargets(subDir, targets) == nil {
+				step.Status = "completed"
+				step.Error = ""
+				st.Plan.Workflow.Steps[i] = step
+				st.mu.Lock()
+				_ = mergeStepArtifacts(&st.BB, subDir)
+				err := st.save()
+				st.mu.Unlock()
+				if err != nil {
+					return nil, err
+				}
+				drv := &flowDriver{opts: e.deps}
+				drv.emit(st, Event{Type: EvStepCompleted, StepID: step.ID, Role: step.Role,
+					Text: fmt.Sprintf("中断恢复：产物齐全，按 attempt %s 判定完成", step.AttemptID)})
+				continue
+			}
+			st.Plan.Workflow.Steps[i].Status = "pending"
+			if err := st.save(); err != nil {
+				return nil, err
+			}
+			return nil, nil
+		default:
+			drv := &flowDriver{opts: e.deps}
 			out, _, err := drv.hold(st, HitlConfirmation, step.ID, step.Role,
-				fmt.Sprintf("步骤「%s」在执行中中断，且没有可信任的完成记录。%s 如何继续？", step.Title, reason),
+				fmt.Sprintf("步骤「%s」在执行中中断。如何继续？", step.Title),
 				[]HumanOption{
-					{Value: "retry", Label: "明确重试（接收可能的副作用）"},
+					{Value: "retry", Label: "重试"},
 					{Value: "skip", Label: "跳过该步骤"},
 					{Value: "abort", Label: "中止运行"},
 				},
 				"interrupted", "interrupted", nil)
 			return out, err
-		default:
-			now := time.Now()
-			st.Meta.Status = string(RunWaitingReconciliation)
-			st.Meta.UpdatedAt = now
-			if err := st.save(); err != nil {
-				return nil, err
-			}
-			drv.emit(st, Event{Type: EvPhase, Text: fmt.Sprintf("中断恢复：%s进入 reconciliation（reason=%s）", step.ID, reason)})
-			return &RunResult{
-				RunID:   st.Meta.ID,
-				Status:  RunWaitingReconciliation,
-				Plan:    planPtr(st),
-				WorkDir: st.Meta.WorkDir,
-				Summary: "interrupted-running step requires reconciliation",
-				Error:   reason,
-			}, nil
 		}
 	}
 	return nil, nil
@@ -1217,13 +1143,6 @@ func (e *SquadEngine) Resume(ctx context.Context, runID string) (*RunResult, err
 	if RunStatus(st.Meta.Status).Terminal() {
 		return e.Inspect(runID)
 	}
-	// F05/U3b：中断恢复产出 reconciliation 状态后，后续 Resume 只报告现状，不
-	// 继续推进；除非用户显式响应（TODO: 对应节，又意图决定在能
-	// 走 steer/respond/recovery后被 reconciliation_action 处理；本轮仅表
-	// 现状，不隐默重跑）。
-	if RunStatus(st.Meta.Status) == RunWaitingReconciliation {
-		return e.Inspect(runID)
-	}
 	// 图连续性校验（Q1 WP4/WP1 裁定）：崩溃期间的检查点图指纹与当前 Steps
 	// 重编译不一致 = 图形态被改过（动态调整或外力），拒绝恢复并提示重新规划。
 	if err := validateGraphContinuity(st); err != nil {

squad/executor_result_contract_test.go

@@ -1,517 +0,0 @@
-// Package squad: F05/U3b executor result 合约测试。
-//
-// 覆盖以下消费规则：
-//   - (nil,nil) executor 返回 (result==nil && err==nil) 是稳定失败（非 panic，非 success）。
-//   - Success=true 但无执行证据（patches-only）也要走交付物校验；
-//     失败-with-full-patches 仍为失败（不允许文件存在翻成 success）。
-//   - canceled/aborted 结果不进入 retry 通路（不是业务失败，不可重试）。
-//   - side_effects=possible 的失败禁止自动重试；
-//     side_effects=none 的纯读取失败可走有界重试。
-//   - recovery：trusted record 存在 + 文件缺失 -> 恢复 completed（权威来源 = record）；
-//     无 record + 文件齐全 -> needs_reconciliation（不假装完成、不隐默 rerun）；
-//     无 record + 文件缺失 -> needs_reconciliation；
-//     record SideEffects=none + 缺失文件 -> 一次重试 pending（不重复执行副作用）。
-//
-// 这些测试使用 squad 包的 in-memory executor 替身进行 hermetic 验证；不调用
-// 真实模型或外部组件。
-package squad
-
-import (
-	"context"
-	"os"
-	"path/filepath"
-	"strings"
-	"sync"
-	"sync/atomic"
-	"testing"
-	"time"
-)
-
-// ---- 替身 executor ----
-
-// frozenResultBackend 每次 Execute 返回同一个固定结果（用于直接构造语义场景）。
-type frozenResultBackend struct {
-	mu    sync.Mutex
-	calls int32
-	res   *ExecResult
-	err   error
-}
-
-func (b *frozenResultBackend) Name() string { return "frozen" }
-
-func (b *frozenResultBackend) Execute(_ context.Context, _ ExecRequest) (*ExecResult, error) {
-	atomic.AddInt32(&b.calls, 1)
-	b.mu.Lock()
-	defer b.mu.Unlock()
-	if b.err != nil {
-		return b.res, b.err
-	}
-	return b.res, nil
-}
-
-func (b *frozenResultBackend) callCount() int32 {
-	return atomic.LoadInt32(&b.calls)
-}
-
-// materializingBackend 每次 Execute 写入 patches 但把 Success 留给 caller 决定。
-// 用于「patches-only success」测试场景。
-type materializingBackend struct {
-	mu    sync.Mutex
-	calls int32
-	// success 只在 success==true 时返回 Success=true；测试 override 用。
-	success  bool
-	canceled bool
-	aborted  bool
-	// sideEffects 决定是否允许重试。
-	sideEffects ExecSideEffects
-	// writePatch 控制是否 materializes patch 到 workDir。
-	writePatch bool
-}
-
-func (b *materializingBackend) Name() string { return "materializing" }
-
-func (b *materializingBackend) Execute(_ context.Context, req ExecRequest) (*ExecResult, error) {
-	atomic.AddInt32(&b.calls, 1)
-	patches := map[string][]byte{}
-	for _, t := range req.Targets {
-		if b.writePatch {
-			_ = os.WriteFile(filepath.Join(req.WorkDir, t), []byte(goodReport), 0o644)
-		}
-		patches[t] = []byte(goodReport)
-	}
-	return &ExecResult{
-		Success:     b.success,
-		Patches:     patches,
-		Logs:        "stub",
-		SideEffects: b.sideEffects,
-		Canceled:    b.canceled,
-		Aborted:     b.aborted,
-	}, nil
-}
-
-func (b *materializingBackend) callCount() int32 {
-	return atomic.LoadInt32(&b.calls)
-}
-
-// ---- 测试辅助 ----
-
-func newFrozenTestEngine(t *testing.T, dir string, backend Executor) *SquadEngine {
-	t.Helper()
-	reg := NewRoleRegistry()
-	if err := reg.Register(RoleSpec{Role: "researcher", Name: "调研", Backend: backend.Name()}); err != nil {
-		t.Fatal(err)
-	}
-	if err := reg.Register(RoleSpec{Role: "writer", Name: "撰写", Backend: backend.Name()}); err != nil {
-		t.Fatal(err)
-	}
-	if err := reg.Register(RoleSpec{Role: "reviewer", Name: "审核", Backend: backend.Name()}); err != nil {
-		t.Fatal(err)
-	}
-	backends := NewBackendResolver()
-	backends.RegisterInstance(backend.Name(), backend)
-	eng, err := NewEngine(Options{
-		DataDir: dir, Registry: reg, Planner: mockScriptPlanner(baseSteps()...),
-		Backends: backends, Sink: NopSink{},
-	})
-	if err != nil {
-		t.Fatal(err)
-	}
-	return eng
-}
-
-// ---- 合约测试 ----
-
-// TestNilNilResultIsStableFailure：(nil,nil) 必须稳定失败——不 panic、不被 papered over。
-func TestNilNilResultIsStableFailure(t *testing.T) {
-	dir := t.TempDir()
-	be := &frozenResultBackend{res: nil, err: nil}
-	eng := newFrozenTestEngine(t, dir, be)
-	res, err := eng.Run(context.Background(), Goal{Text: "空结果"}, RunOptions{})
-	if err != nil {
-		t.Fatalf("run returned engine error: %v", err)
-	}
-	// 关键失败 → RunFailed；步骤状态 failed；call 一次；不 panic。
-	if res.Status != RunFailed {
-		t.Fatalf("want RunFailed, got %s (err=%s)", res.Status, res.Error)
-	}
-	if be.callCount() != 1 {
-		t.Fatalf("backend called %d times, want 1", be.callCount())
-	}
-	if res.Plan == nil {
-		t.Fatal("plan missing")
-	}
-	if got := res.Plan.Steps[0].Status; got != "failed" {
-		t.Fatalf("step 1 status = %s, want failed", got)
-	}
-	if got := res.Plan.Steps[0].Error; got != "executor returned nil result" {
-		t.Fatalf("step 1 error = %q, want executor-nil-result reason", got)
-	}
-}
-
-// TestSuccessWithFullPatchesMaterializes：Success=true + 已有 patch
-// —— 通过交付物校验完成；patches-only success 也走同一路径。
-func TestSuccessWithFullPatchesMaterializes(t *testing.T) {
-	dir := t.TempDir()
-	be := &materializingBackend{success: true, sideEffects: ExecSideEffectsNone, writePatch: true}
-	eng := newFrozenTestEngine(t, dir, be)
-	res, err := eng.Run(context.Background(), Goal{Text: "patches 落地校验"}, RunOptions{})
-	if err != nil {
-		t.Fatalf("run: %v", err)
-	}
-	if res.Status != RunCompleted {
-		t.Fatalf("want completed, got %s err=%s", res.Status, res.Error)
-	}
-	// 三步各调用一次；patches 命中产物路径且 Success=true，每步一次。
-	if got := be.callCount(); got < 3 {
-		t.Fatalf("backend called %d times, want at least 3 (3 steps)", got)
-	}
-}
-
-// TestFailedWithFullPatchesStaysFailed：F05/KTD4 关键不变式：失败 + 文件齐全
-// 也不能翻成成功——即便 patches 内容已落地。
-func TestFailedWithFullPatchesStaysFailed(t *testing.T) {
-	dir := t.TempDir()
-	be := &materializingBackend{success: false, sideEffects: ExecSideEffectsPossible, writePatch: true}
-	eng := newFrozenTestEngine(t, dir, be)
-	res, err := eng.Run(context.Background(), Goal{Text: "失败-but-files-present"}, RunOptions{})
-	if err != nil {
-		t.Fatalf("run: %v", err)
-	}
-	if res.Status != RunFailed {
-		t.Fatalf("want RunFailed, got %s", res.Status)
-	}
-	if res.Plan == nil || res.Plan.Steps[0].Status != "failed" {
-		t.Fatalf("step 1 should be failed, plan=%+v", res.Plan)
-	}
-	// 关键：失败被 failStep 重判后，不应回写 completed；也不应自动 rerun。
-	// SideEffects=possible 禁止重试：s1 仅调用一次，后续步骤不再调用（运行已中止）。
-	if got := be.callCount(); got > 1 {
-		t.Fatalf("failed step retried/executed again: calls=%d", got)
-	}
-}
-
-// TestCanceledResultIsNotRetried：canceled 结果不允许自动重试（KTD4）：
-// canceled 是显式意图，不是 provider 瞬时故障。
-func TestCanceledResultIsNotRetried(t *testing.T) {
-	dir := t.TempDir()
-	be := &materializingBackend{success: false, canceled: true, sideEffects: ExecSideEffectsNone, writePatch: false}
-	eng := newFrozenTestEngine(t, dir, be)
-	res, err := eng.Run(context.Background(), Goal{Text: "取消不可重试"}, RunOptions{})
-	if err != nil {
-		t.Fatalf("run: %v", err)
-	}
-	// 失败（不是 panic），且仅被调用一次（不允许 retry）。
-	if res.Status != RunFailed {
-		t.Fatalf("want RunFailed, got %s", res.Status)
-	}
-	if be.callCount() != 1 {
-		t.Fatalf("canceled invoked %d times, want 1 (no retry)", be.callCount())
-	}
-}
-
-// TestAbortedResultIsNotRetried：aborted 同 canceled 语义。
-func TestAbortedResultIsNotRetried(t *testing.T) {
-	dir := t.TempDir()
-	be := &materializingBackend{success: false, aborted: true, sideEffects: ExecSideEffectsNone, writePatch: false}
-	eng := newFrozenTestEngine(t, dir, be)
-	res, err := eng.Run(context.Background(), Goal{Text: "中止不可重试"}, RunOptions{})
-	if err != nil {
-		t.Fatalf("run: %v", err)
-	}
-	if res.Status != RunFailed {
-		t.Fatalf("want RunFailed, got %s", res.Status)
-	}
-	if be.callCount() != 1 {
-		t.Fatalf("aborted invoked %d times, want 1 (no retry)", be.callCount())
-	}
-}
-
-// TestSideEffectsPossibleNotRetried：side_effects=possible 失败禁止自动重试
-// （防止重放已发生的外部动作）。
-func TestSideEffectsPossibleNotRetried(t *testing.T) {
-	dir := t.TempDir()
-	be := &materializingBackend{success: false, sideEffects: ExecSideEffectsPossible, writePatch: false}
-	eng := newFrozenTestEngine(t, dir, be)
-	res, err := eng.Run(context.Background(), Goal{Text: "可能副作用失败不重试"}, RunOptions{})
-	if err != nil {
-		t.Fatalf("run: %v", err)
-	}
-	if res.Status != RunFailed {
-		t.Fatalf("want RunFailed, got %s", res.Status)
-	}
-	if be.callCount() != 1 {
-		t.Fatalf("possible-side-effects invoked %d times, want 1 (no retry)", be.callCount())
-	}
-}
-
-// TestSideEffectsNoneRetriedWithinBounds：side_effects=none 的失败保留
-// 有界重试通路（最多 MaxTransientStepInvocations 次），不会跨步骤。
-func TestSideEffectsNoneRetriedWithinBounds(t *testing.T) {
-	dir := t.TempDir()
-	// stage 1: failed with side_effects=none → 触发重试；
-	// stage 2: 返回 success。后续调用返回 last state（success）。
-	be := &stagedBackend{states: []*ExecResult{
-		{Success: false, SideEffects: ExecSideEffectsNone, ErrorReason: "transient"},
-		{Success: true, SideEffects: ExecSideEffectsNone},
-	}}
-	eng := newFrozenTestEngine(t, dir, be)
-	res, err := eng.Run(context.Background(), Goal{Text: "可重试"}, RunOptions{})
-	if err != nil {
-		t.Fatalf("run: %v", err)
-	}
-	if res.Status != RunCompleted {
-		t.Fatalf("want RunCompleted, got %s err=%s", res.Status, res.Error)
-	}
-	// 全运行计数：s1 经历重试（最多 2 次调用）；后续步骤首次成功。
-	// 说明 重试仅限于 s1，不超出单步边界。
-	if be.callCount() < 2 {
-		t.Fatalf("retried count = %d, want at least 2 (failed then success on s1)", be.callCount())
-	}
-}
-
-// ---- 恢复 / 对账 ----
-
-// TestRecoveryTrustedRecordCompletesMissingFiles：F05 关键场景：trusted terminal
-// record 存在 + 文件被外部删 → 仍恢复为 completed（state 是权威，files 仅诊断）。
-func TestRecoveryTrustedRecordCompletesMissingFiles(t *testing.T) {
-	backend := &perStepRecordingBackend{}
-	dir := t.TempDir()
-	eng := newAttemptTestEngine(t, backend, dir)
-	res, err := eng.Run(context.Background(), Goal{Text: "record-only-path"}, RunOptions{})
-	if err != nil || res.Status != RunCompleted {
-		t.Fatalf("setup run: %v %+v", err, res)
-	}
-	// 准备恢复场景：删除 s2 的 output.md，但保留 trusted record。
-	st, err := loadRunState(res.WorkDir)
-	if err != nil {
-		t.Fatal(err)
-	}
-	sub := filepath.FromSlash(st.Plan.Workflow.Steps[1].SubWorkspace)
-	if err := os.Remove(filepath.Join(res.WorkDir, sub, "output.md")); err != nil {
-		t.Fatal(err)
-	}
-	// 记录已写入（setup 步骤会触发 recordStepTerminal）。
-	terminals := loadStepTerminalsForWorkDir(res.WorkDir)
-	if _, ok := terminals[st.Plan.Workflow.Steps[1].ID]; !ok {
-		t.Fatalf("trusted record missing for %s", st.Plan.Workflow.Steps[1].ID)
-	}
-	// Flip s2 to running（带 record）。
-	st2, _ := loadRunState(res.WorkDir)
-	st2.Plan.Workflow.Steps[1].Status = "running"
-	st2.Plan.Workflow.Steps[1].Attempts = 2
-	st2.setWorkflow(st2.Plan.Workflow)
-	if err := st2.save(); err != nil {
-		t.Fatal(err)
-	}
-	eng2 := newAttemptTestEngine(t, backend, dir)
-	res2, err := eng2.Resume(context.Background(), res.RunID)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if res2.Status != RunCompleted {
-		t.Fatalf("trusted record should restore completed, got %s", res2.Status)
-	}
-}
-
-// TestRecoveryNoRecordWithFilesNeedsReconciliation：F05：trusted record 缺失但
-// 文件齐全 → 不允许隐默 rerun 也不允许假装完成；必须进 reconciliation。
-func TestRecoveryNoRecordWithFilesNeedsReconciliation(t *testing.T) {
-	backend := &perStepRecordingBackend{}
-	dir := t.TempDir()
-	eng := newAttemptTestEngine(t, backend, dir)
-	res, err := eng.Run(context.Background(), Goal{Text: "no-record-with-files", HitlMode: HitlAuto}, RunOptions{})
-	if err != nil || res.Status != RunCompleted {
-		t.Fatalf("setup run: %v %+v", err, res)
-	}
-	flipToInterrupted(t, res.WorkDir, 1)
-	// flipToInterrupted 现在已经删除 record；显式再校验一次。
-	ws := res.WorkDir
-	terminals := loadStepTerminalsForWorkDir(ws)
-	if _, ok := terminals["s2"]; ok {
-		t.Fatalf("trusted record for s2 should be removed")
-	}
-	eng2 := newAttemptTestEngine(t, backend, dir)
-	res2, err := eng2.Resume(context.Background(), res.RunID)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if res2.Status != RunWaitingReconciliation {
-		t.Fatalf("want waiting_reconciliation, got %s", res2.Status)
-	}
-	if res2.Plan == nil {
-		t.Fatal("plan missing")
-	}
-	var recon int
-	for _, s := range res2.Plan.Steps {
-		if s.ID == "s2" && s.Status == StepStatusNeedsReconciliation {
-			recon++
-		}
-	}
-	if recon != 1 {
-		t.Fatalf("s2 should be needs_reconciliation (plan=%+v)", res2.Plan.Steps)
-	}
-}
-
-// TestRecoveryNoRecordMissingFilesNeedsReconciliation：F05：trusted record 缺失
-// 且文件缺失 → 必须进 reconciliation。
-func TestRecoveryNoRecordMissingFilesNeedsReconciliation(t *testing.T) {
-	backend := &perStepRecordingBackend{}
-	dir := t.TempDir()
-	eng := newAttemptTestEngine(t, backend, dir)
-	res, err := eng.Run(context.Background(), Goal{Text: "no-record-missing-files", HitlMode: HitlAuto}, RunOptions{})
-	if err != nil || res.Status != RunCompleted {
-		t.Fatalf("setup run: %v %+v", err, res)
-	}
-	flipToInterrupted(t, res.WorkDir, 1)
-	st, _ := loadRunState(res.WorkDir)
-	sub := filepath.FromSlash(st.Plan.Workflow.Steps[1].SubWorkspace)
-	if err := os.Remove(filepath.Join(res.WorkDir, sub, "output.md")); err != nil {
-		t.Fatal(err)
-	}
-	eng2 := newAttemptTestEngine(t, backend, dir)
-	res2, err := eng2.Resume(context.Background(), res.RunID)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if res2.Status != RunWaitingReconciliation {
-		t.Fatalf("want waiting_reconciliation, got %s", res2.Status)
-	}
-}
-
-// TestRetryReusesExecutionIDAndDeadline：F05/KTD4：重试必须复用同一 ExecutionID
-// 与同一 Deadline（防止下游幂等键漂移 + budget 重新计票）。
-func TestRetryReusesExecutionIDAndDeadline(t *testing.T) {
-	capturer := &executionIDCapturingBackend{}
-	dir := t.TempDir()
-	reg := NewRoleRegistry()
-	if err := reg.Register(RoleSpec{Role: "researcher", Name: "调研", Backend: "capture"}); err != nil {
-		t.Fatal(err)
-	}
-	if err := reg.Register(RoleSpec{Role: "writer", Name: "撰写", Backend: "capture"}); err != nil {
-		t.Fatal(err)
-	}
-	if err := reg.Register(RoleSpec{Role: "reviewer", Name: "审核", Backend: "capture"}); err != nil {
-		t.Fatal(err)
-	}
-	backends := NewBackendResolver()
-	backends.RegisterInstance("capture", capturer)
-	eng, err := NewEngine(Options{
-		DataDir: dir, Registry: reg, Planner: mockScriptPlanner(baseSteps()...),
-		Backends: backends, Sink: NopSink{},
-	})
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, err := eng.Run(context.Background(), Goal{Text: "retry-id-stable"}, RunOptions{})
-	if err != nil {
-		t.Fatalf("run: %v", err)
-	}
-	if res.Status != RunCompleted {
-		t.Fatalf("want completed, got %s err=%s", res.Status, res.Error)
-	}
-	if len(capturer.calls) == 0 {
-		t.Fatal("no calls captured")
-	}
-	// ExecutionID 格式 runID/stepID/N；同一步的所有调用，第二段必为同一 stepID。
-	// 我们验证「同一 stepID 的不同调用共享同一 ExecutionID 与 Deadline」。
-	byStep := map[string][]capturedCall{}
-	for _, c := range capturer.calls {
-		parts := strings.Split(c.executionID, "/")
-		if len(parts) < 2 {
-			continue
-		}
-		stepID := parts[1]
-		byStep[stepID] = append(byStep[stepID], c)
-	}
-	if len(byStep["s1"]) == 0 {
-		t.Fatalf("no s1 calls captured: %+v", capturer.calls)
-	}
-	if len(byStep["s2"]) == 0 {
-		t.Fatalf("no s2 calls captured")
-	}
-	// 检查每一步至少被调用一次；后续步骤也应该被调用。
-	for _, id := range []string{"s1", "s2", "s3"} {
-		if cs := byStep[id]; len(cs) == 0 {
-			t.Fatalf("step %s not captured", id)
-		}
-	}
-}
-
-// executionIDCapturingBackend 记录每一次调用的 ExecutionID/Deadline。
-type executionIDCapturingBackend struct {
-	mu    sync.Mutex
-	calls []capturedCall
-}
-
-type capturedCall struct {
-	stepID      string
-	executionID string
-	deadline    string
-}
-
-func (b *executionIDCapturingBackend) Name() string { return "capture" }
-
-func (b *executionIDCapturingBackend) Execute(_ context.Context, req ExecRequest) (*ExecResult, error) {
-	b.mu.Lock()
-	defer b.mu.Unlock()
-	deadline := ""
-	if req.Budget != nil {
-		deadline = req.Budget.Deadline.Format(time.RFC3339Nano)
-	}
-	b.calls = append(b.calls, capturedCall{
-		stepID:      filepath.Base(req.WorkDir),
-		executionID: req.ExecutionID,
-		deadline:    deadline,
-	})
-	patches := make(map[string][]byte, len(req.Targets))
-	for _, t := range req.Targets {
-		patches[t] = []byte(goodReport)
-		_ = os.WriteFile(filepath.Join(req.WorkDir, t), []byte(goodReport), 0o644)
-	}
-	return &ExecResult{
-		Success:     true,
-		Patches:     patches,
-		SideEffects: ExecSideEffectsNone,
-	}, nil
-}
-
-// stagedBackend 按调用顺序返回预置的不同结果（用于「失败→成功」retry 验证）。
-type stagedBackend struct {
-	mu     sync.Mutex
-	states []*ExecResult
-	idx    int32
-}
-
-func (b *stagedBackend) Name() string { return "staged" }
-
-func (b *stagedBackend) Execute(_ context.Context, req ExecRequest) (*ExecResult, error) {
-	n := atomic.AddInt32(&b.idx, 1)
-	b.mu.Lock()
-	idx := int(n) - 1
-	if idx >= len(b.states) {
-		idx = len(b.states) - 1
-	}
-	res := b.states[idx]
-	b.mu.Unlock()
-	if res == nil {
-		return nil, nil
-	}
-	// 输出：填充 patches 以该调用请求的 targets 返回，以便后续验收级联。
-	patches := make(map[string][]byte, len(req.Targets))
-	for _, t := range req.Targets {
-		patches[t] = []byte(goodReport)
-	}
-	out := *res
-	out.Patches = patches
-	if out.Success {
-		// 落盘：让交付物验收能拿到文件。
-		for _, t := range req.Targets {
-			_ = os.WriteFile(filepath.Join(req.WorkDir, t), []byte(goodReport), 0o644)
-		}
-	}
-	return &out, nil
-}
-
-func (b *stagedBackend) callCount() int32 {
-	return atomic.LoadInt32(&b.idx)
-}

squad/groupchat.go

@@ -339,13 +339,7 @@ func (e *SquadEngine) gcFinish(st *RunState, drv *flowDriver) *RunResult {
 	st.Meta.Status = string(RunCompleted)
 	now := time.Now()
 	st.Meta.CompletedAt = &now
-	// F11c：终态落盘先于返回 RunCompleted——写不进去就不能报完成。
-	if serr := st.save(); serr != nil {
-		st.mu.Unlock()
-		msg := persistenceFailureText("save terminal run state", serr)
-		return &RunResult{RunID: st.Meta.ID, Status: RunFailed, Plan: planPtr(st),
-			WorkDir: st.Meta.WorkDir, Error: msg, Summary: msg}
-	}
+	_ = st.save()
 	st.mu.Unlock()
 	summary := fmt.Sprintf("group_chat：%d 轮发言，转录已收口", turns)
 	drv.emit(st, Event{Type: EvPhase, Text: summary})

squad/leader.go

@@ -33,7 +33,7 @@ func leaderExecRequest(drv *flowDriver, st *RunState, leader RoleSpec, workDir,
 		Budget:        budget,
 		OnEvent: func(ev ProcessEvent) {
 			drv.emit(st, Event{Type: EvDetail, StepID: stepID, Role: leader.Role, Text: ev.Text,
-				Data: processEventData(ev)})
+				Data: marshalJSON(map[string]string{"process_type": ev.Type, "tool": ev.Tool, "detail": ev.Detail})})
 		},
 	}
 }
@@ -301,38 +301,12 @@ func (e *SquadEngine) runLeaderLoop(ctx context.Context, st *RunState) (*RunResu
 		prompt = st.injectSteering(prompt)
 		instructions := readFileOr(filepath.Join(subDir, "AGENTS.md"), leaderPrompt(members, e.opts.IntentCatalog.intentCatalogMd()))
 		result, execErr := backend.Execute(ctx, leaderExecRequest(drv, st, leader, subDir, prompt, instructions, []string{"delegation.json"}, e.deps.StepTimeout))
-		// F05/KTD4：与 runStep 保持同一语义——(nil,nil)、Canceled、文件存在都不能翻成 Success。
-		if execErr != nil && result == nil {
-			_ = os.Remove(filepath.Join(subDir, "delegation.json"))
-			return e.leaderFail(st, fmt.Sprintf("leader turn %d: executor returned no result and no error (backend=%s)", st.Leader.Turns, backendName)), nil, nil
-		}
-		if result == nil {
-			_ = os.Remove(filepath.Join(subDir, "delegation.json"))
-			return e.leaderFail(st, fmt.Sprintf("leader turn %d: executor returned nil result (backend=%s)", st.Leader.Turns, backendName)), nil, nil
-		}
-		if result.IsCanceled() {
-			reason := result.ErrorReason
-			if reason == "" {
-				reason = "executor canceled"
-			}
-			_ = os.Remove(filepath.Join(subDir, "delegation.json"))
-			return e.leaderFail(st, fmt.Sprintf("leader turn %d: %s (backend=%s)", st.Leader.Turns, reason, backendName)), nil, nil
-		}
-		// F05/KTD4：不允许「文件已存在」替失败翻成成功；仅当 result.Success=true 且
-		// delegation.json 已物化时进入下方解析路径。patches-only success 也走同样的
-		// 解析路径（不绕过）。
-		delegationMissing := false
-		if _, statErr := os.Stat(filepath.Join(subDir, "delegation.json")); statErr != nil {
-			delegationMissing = true
-		}
-		if execErr != nil || !result.Success || delegationMissing {
+		if execErr != nil || result == nil || !result.Success {
 			reason := "leader execution failed"
 			if execErr != nil {
 				reason = execErr.Error()
-			} else if result.ErrorReason != "" {
+			} else if result != nil && result.ErrorReason != "" {
 				reason = result.ErrorReason
-			} else if delegationMissing {
-				reason = "delegation.json not produced despite Success=true"
 			}
 			_ = os.Remove(filepath.Join(subDir, "delegation.json"))
 			return e.leaderFail(st, fmt.Sprintf("leader turn %d: %s (backend=%s)", st.Leader.Turns, reason, backendName)), nil, nil

squad/models.go

@@ -296,21 +296,12 @@ const (
 	RunDegraded              RunStatus = "degraded"
 	RunFailed                RunStatus = "failed"
 	RunCancelled             RunStatus = "cancelled"
-	// RunWaitingReconciliation F05：recovery 期间无法判定步骤结果（无 trusted
-	// terminal record、文件存在性与 record 冲突等）；非交互模式下记录该状态
-	// 而非永久挂起。交互模式下挂起（合成 HITL 门），与既有 HITL 通道同款语义。
-	RunWaitingReconciliation RunStatus = "waiting_reconciliation"
 )
 
-// StepStatusNeedsReconciliation 是步骤级 needs-reconciliation 状态。
-// 仅供 recovery 使用；runStep 不在成功路径写入。
-const StepStatusNeedsReconciliation = "needs_reconciliation"
-
 // IsHeld 返回该状态是否为 HITL 挂起态（等待人工回应）。
 func (s RunStatus) IsHeld() bool {
 	return s == RunAwaitingClarification || s == RunAwaitingConfirmation ||
-		s == RunWaitingHuman || s == RunQualityReview ||
-		s == RunWaitingReconciliation
+		s == RunWaitingHuman || s == RunQualityReview
 }
 
 // Terminal 报告该状态是否为终态。

squad/port.go

@@ -75,32 +75,9 @@ type ProcessEvent struct {
 	LastCompletionTokens int
 	CacheReadTokens      int
 	CacheWriteTokens     int
-	// CallID 是本次工具调用的身份（与 native exec.ProcessEvent.CallID 同语义）：
-	// tool_call_delta / tool_execution_start / tool_execution_end 携带同一 ID，
-	// 消费方按 ID 配对而不是按到达顺序——并行工具调用会交错到达。
-	// 空串 = 上游未填充（legacy 版本），不是「一次无名调用」。
-	CallID string `json:"call_id,omitempty"`
-	// Status 是 tool_execution_end 的真实结局，取值见 ToolStatus*；
-	// 空串 = legacy/未知，不据此判定失败（与既有行为一致）。
-	Status string `json:"status,omitempty"`
-	// ErrorReason 是 Status=error 时的真实失败原因原文（与 native 同语义）；
-	// 空串 = 无失败或上游未填充。
-	ErrorReason string `json:"error_reason,omitempty"`
 }
 
-// ProcessEvent.Status 的取值域（与 native 侧一致；空串 = legacy/未知）。
-const (
-	ToolStatusOK       = "ok"
-	ToolStatusError    = "error"
-	ToolStatusDenied   = "denied"
-	ToolStatusCanceled = "canceled"
-)
-
 // ExecResult 是一次步骤执行的终态。
-//
-// 语义契约（F05/U3b）：Success=true 才是经过验证的成功；仅在交付物齐全且
-// ExecutionID 复用前提下成立。仅凭已物化的文件存在不能把失败翻成成功。
-// 后端可以选择性填写以下可选字段；缺省 = 旧的读取行为。
 type ExecResult struct {
 	Success          bool              `json:"success"`
 	Patches          map[string][]byte `json:"patches"`
@@ -109,36 +86,6 @@ type ExecResult struct {
 	Retryable        bool              `json:"retryable,omitempty"`
 	PromptTokens     int               `json:"prompt_tokens,omitempty"`
 	CompletionTokens int               `json:"completion_tokens,omitempty"`
-	// Canceled 为 true 时表示执行是被请求/中断终止的，不是业务失败。
-	// F05/KTD4：取消结果不应被 retry 改判为可重试；这是用户/系统显式意图。
-	Canceled bool `json:"canceled,omitempty"`
-	// Aborted 与 Canceled 同语义但语义更具体（异常终止/被调用方 abort）。
-	Aborted bool `json:"aborted,omitempty"`
-	// SideEffects 报告执行是否可能产生过外部动作。空值 = "possible"（旧行为）。
-	//   none     = 确认无副作用（未派发或纯读取）——可重试
-	//   possible = 可能产生过副作用（文件已写/外部动作已发生）——禁止无条件重试
-	SideEffects ExecSideEffects `json:"side_effects,omitempty"`
-}
-
-// ExecSideEffects 报告执行是否可能产生过外部副作用；缺席 = possible（保守）。
-type ExecSideEffects string
-
-const (
-	// ExecSideEffectsNone 确认未产生任何外部副作用；可重试。
-	ExecSideEffectsNone ExecSideEffects = "none"
-	// ExecSideEffectsPossible 可能产生过副作用；禁止无条件重试。
-	ExecSideEffectsPossible ExecSideEffects = "possible"
-)
-
-// IsRetryableSideEffectReports 报告 SideEffects 是否为「可重试」形态。
-// 默认 possible 视为不可重试（保守）：调用方必须显式声明 none 才放开重试。
-func (r *ExecResult) IsRetryableSideEffectReports() bool {
-	return r != nil && r.SideEffects == ExecSideEffectsNone
-}
-
-// IsCanceled 报告结果是否被取消/中止（不是业务失败）。
-func (r *ExecResult) IsCanceled() bool {
-	return r != nil && (r.Canceled || r.Aborted)
 }
 
 // StubExecutor 是离线桩，测试与未接模型时使用。

squad/process_event_test.go

@@ -1,222 +0,0 @@
-package squad
-
-import (
-	"context"
-	"encoding/json"
-	"fmt"
-	"strings"
-	"sync"
-	"testing"
-)
-
-const toolCallFailure = "grep: permission denied"
-
-// concurrentToolBackend 在一次执行里并发发两次同名工具调用（start + end）：
-// CallID 每次执行唯一、一成一败。用于验证过程帧经扇出后仍按 CallID 配对、
-// 结局不串（并行工具调用交错到达时不再靠到达顺序猜）。
-type concurrentToolBackend struct {
-	inner Executor
-
-	mu   sync.Mutex
-	exec int
-}
-
-func (b *concurrentToolBackend) Name() string { return "concurrent-tool" }
-
-func (b *concurrentToolBackend) Execute(ctx context.Context, req ExecRequest) (*ExecResult, error) {
-	b.mu.Lock()
-	b.exec++
-	round := b.exec
-	b.mu.Unlock()
-
-	if req.OnEvent != nil {
-		var wg sync.WaitGroup
-		for _, suffix := range []string{"ok", "err"} {
-			wg.Add(1)
-			go func(suffix string) {
-				defer wg.Done()
-				callID := fmt.Sprintf("call_%d_%s", round, suffix)
-				req.OnEvent(ProcessEvent{Type: "tool_execution_start", Tool: "grep",
-					CallID: callID, Detail: `{"q":"x"}`})
-				status, reason, text := ToolStatusOK, "", "found"
-				if suffix == "err" {
-					status, reason, text = ToolStatusError, toolCallFailure, ""
-				}
-				req.OnEvent(ProcessEvent{Type: "tool_execution_end", Tool: "grep", CallID: callID,
-					Status: status, ErrorReason: reason, Text: text})
-			}(suffix)
-		}
-		wg.Wait() // 后端自己等待：并行工具调用在同一执行内收敛
-	}
-	if b.inner == nil {
-		return &ExecResult{Success: true}, nil
-	}
-	return b.inner.Execute(ctx, req)
-}
-
-// detailFrame 是 detail 事件 Data 的读取视图。
-type detailFrame struct {
-	processType string
-	tool        string
-	callID      string
-	status      string
-	errorReason string
-}
-
-func readDetailFrames(t *testing.T, events []Event) []detailFrame {
-	t.Helper()
-	var out []detailFrame
-	for _, ev := range events {
-		if ev.Type != EvDetail {
-			continue
-		}
-		var data map[string]string
-		if err := json.Unmarshal(ev.Data, &data); err != nil {
-			t.Fatalf("detail data must stay a JSON object: %v (%s)", err, ev.Data)
-		}
-		out = append(out, detailFrame{
-			processType: data["process_type"], tool: data["tool"], callID: data["call_id"],
-			status: data["status"], errorReason: data["error_reason"],
-		})
-	}
-	return out
-}
-
-// assertToolCallsPairedByCallID 断言并发同名工具调用在 detail 事件流里按 CallID
-// 配对：每个 call_id 恰好一帧 start、一帧 end，结局与该调用自身一致。
-func assertToolCallsPairedByCallID(t *testing.T, frames []detailFrame) {
-	t.Helper()
-	starts := map[string]int{}
-	ends := map[string]detailFrame{}
-	for _, f := range frames {
-		if f.callID == "" {
-			t.Fatalf("frame lost its CallID: %+v", f)
-		}
-		if f.tool != "grep" {
-			t.Fatalf("frame tool = %q, want grep", f.tool)
-		}
-		switch f.processType {
-		case "tool_execution_start":
-			starts[f.callID]++
-		case "tool_execution_end":
-			if _, dup := ends[f.callID]; dup {
-				t.Fatalf("duplicate end frame for %s: %+v", f.callID, f)
-			}
-			ends[f.callID] = f
-		}
-	}
-	if len(ends) == 0 {
-		t.Fatalf("no tool_execution_end frame reached the detail stream: %+v", frames)
-	}
-	if len(ends) != len(starts) {
-		t.Fatalf("starts (%d) and ends (%d) must pair by CallID: %+v", len(starts), len(ends), frames)
-	}
-	for callID, end := range ends {
-		if starts[callID] != 1 {
-			t.Fatalf("call %s start frames = %d, want 1", callID, starts[callID])
-		}
-		wantStatus, wantReason := ToolStatusOK, ""
-		if strings.HasSuffix(callID, "_err") {
-			wantStatus, wantReason = ToolStatusError, toolCallFailure
-		}
-		if end.status != wantStatus || end.errorReason != wantReason {
-			t.Fatalf("call %s outcome = %s/%q, want %s/%q（结局必须属于本次调用）",
-				callID, end.status, end.errorReason, wantStatus, wantReason)
-		}
-	}
-}
-
-func TestProcessEventDataLegacyProjectionUnchanged(t *testing.T) {
-	// legacy 上游（无 CallID/Status/ErrorReason）的投影必须逐字节不变。
-	legacy := ProcessEvent{Type: "tool_execution_end", Tool: "grep", Detail: `{"q":"x"}`, Text: "found"}
-	want := `{"detail":"{\"q\":\"x\"}","process_type":"tool_execution_end","tool":"grep"}`
-	if got := string(processEventData(legacy)); got != want {
-		t.Fatalf("legacy detail projection changed:\n got %s\nwant %s", got, want)
-	}
-	// 新字段非空时才出现，既有键仍在。
-	withID := legacy
-	withID.CallID, withID.Status, withID.ErrorReason = "call_1", ToolStatusOK, ""
-	var data map[string]string
-	if err := json.Unmarshal(processEventData(withID), &data); err != nil {
-		t.Fatal(err)
-	}
-	if data["call_id"] != "call_1" || data["status"] != ToolStatusOK {
-		t.Fatalf("identity/outcome must reach the detail stream: %+v", data)
-	}
-	if _, present := data["error_reason"]; present {
-		t.Fatalf("empty outcome must not add a key: %+v", data)
-	}
-	if data["process_type"] != "tool_execution_end" || data["tool"] != "grep" || data["detail"] != `{"q":"x"}` {
-		t.Fatalf("existing projection keys changed: %+v", data)
-	}
-}
-
-func TestStepDetailStreamCarriesConcurrentToolCallIdentity(t *testing.T) {
-	script := map[string]map[string]string{
-		"01_researcher": {"notes.md": "# 笔记\n内容"},
-		"02_writer":     {"output.md": goodReport},
-		"03_reviewer":   {"output.md": goodReport},
-	}
-	dir := t.TempDir()
-	reg := NewRoleRegistry()
-	for _, spec := range []RoleSpec{
-		{Role: "researcher", Name: "调研", Backend: "concurrent-tool"},
-		{Role: "writer", Name: "撰写", Backend: "concurrent-tool"},
-		{Role: "reviewer", Name: "审核", Backend: "concurrent-tool"},
-	} {
-		if err := reg.Register(spec); err != nil {
-			t.Fatal(err)
-		}
-	}
-	backends := NewBackendResolver()
-	backends.RegisterInstance("concurrent-tool", &concurrentToolBackend{
-		inner: &scriptBackend{script: script},
-	})
-	collector := &sliceSink{}
-	engine, err := NewEngine(Options{DataDir: dir, Registry: reg, Planner: mockScriptPlanner(baseSteps()...),
-		Backends: backends, Sink: collector})
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, err := engine.Run(context.Background(), Goal{Text: "生成测试报告"}, RunOptions{})
-	if err != nil {
-		t.Fatal(err)
-	}
-	if res.Status != RunCompleted {
-		t.Fatalf("want completed, got %s (err=%s)", res.Status, res.Error)
-	}
-	assertToolCallsPairedByCallID(t, readDetailFrames(t, collector.events))
-}
-
-func TestLeaderDetailStreamCarriesConcurrentToolCallIdentity(t *testing.T) {
-	inner := &leaderScriptBackend{
-		decisions: []string{delegateJSON("writer", "撰写"), closeJSON},
-		script:    map[string]map[string]string{"01_writer": {"output.md": goodReport}},
-	}
-	dir := t.TempDir()
-	reg := NewRoleRegistry()
-	for _, spec := range []RoleSpec{
-		{Role: "leader", Name: "主管", Backend: "concurrent-tool"},
-		{Role: "writer", Name: "撰写", Backend: "concurrent-tool"},
-	} {
-		if err := reg.Register(spec); err != nil {
-			t.Fatal(err)
-		}
-	}
-	backends := NewBackendResolver()
-	backends.RegisterInstance("concurrent-tool", &concurrentToolBackend{inner: inner})
-	collector := &sliceSink{}
-	engine, err := NewEngine(Options{DataDir: dir, Registry: reg, Backends: backends, Sink: collector})
-	if err != nil {
-		t.Fatal(err)
-	}
-	res, err := engine.Run(context.Background(), Goal{Text: "产出简报"},
-		RunOptions{Mode: "leader", LeaderRole: "leader"})
-	if err != nil {
-		t.Fatal(err)
-	}
-	if res.Status != RunCompleted {
-		t.Fatalf("want completed, got %s err=%s", res.Status, res.Error)
-	}
-	assertToolCallsPairedByCallID(t, readDetailFrames(t, collector.events))
-}

squad/progress.go

@@ -136,22 +136,3 @@ func (s *FanoutSink) Snapshot() ([]Event, int) {
 type NopSink struct{}
 
 func (NopSink) Emit(Event) {}
-
-// processEventData 把过程帧投影成 detail 事件的 Data（JSON 对象）。
-//
-// 工具调用身份与真实结局（call_id / status / error_reason）非空时随帧透传，
-// 使 detail 事件流能按 CallID 配对交错的并行工具调用，而不是按到达顺序猜。
-// 三者全空（legacy 上游未填充）时输出与既有投影逐字节一致，不引入新键。
-func processEventData(ev ProcessEvent) json.RawMessage {
-	data := map[string]string{"process_type": ev.Type, "tool": ev.Tool, "detail": ev.Detail}
-	if ev.CallID != "" {
-		data["call_id"] = ev.CallID
-	}
-	if ev.Status != "" {
-		data["status"] = ev.Status
-	}
-	if ev.ErrorReason != "" {
-		data["error_reason"] = ev.ErrorReason
-	}
-	return marshalJSON(data)
-}

squad/recovery_test.go

@@ -2,7 +2,6 @@ package squad
 
 import (
 	"context"
-	"encoding/json"
 	"errors"
 	"os"
 	"path/filepath"
@@ -163,17 +162,6 @@ func TestInterruptedRunningStepStaysHeld(t *testing.T) {
 	if err := st.save(); err != nil {
 		t.Fatal(err)
 	}
-	// F05/U3b：清除该步 trusted terminal record，模拟“执行中崩溃未达记录点”
-	// recovery 会因为无 record 走 HITL hold 或 reconciliation。
-	terminals := loadStepTerminalsForWorkDir(res.WorkDir)
-	if terminals != nil {
-		stepID := st.Plan.Workflow.Steps[0].ID
-		if _, ok := terminals[stepID]; ok {
-			delete(terminals, stepID)
-			data, _ := json.Marshal(terminals)
-			_ = os.WriteFile(filepath.Join(res.WorkDir, "step_terminals.json"), data, 0o644)
-		}
-	}
 	eng2, err := NewEngine(Options{DataDir: dir, Registry: reg, Planner: eng.opts.Planner, Backends: backends, Sink: NopSink{}})
 	if err != nil {
 		t.Fatal(err)

squad/review.go

@@ -326,13 +326,7 @@ func (e *SquadEngine) runReviewDriver(ctx context.Context, st *RunState) (*RunRe
 	st.Meta.Status = string(status)
 	now := time.Now()
 	st.Meta.CompletedAt = &now
-	// F11c：终态落盘先于返回 RunCompleted——写不进去就不能报完成。
-	if serr := st.save(); serr != nil {
-		st.mu.Unlock()
-		msg := persistenceFailureText("save terminal run state", serr)
-		return &RunResult{RunID: st.Meta.ID, Status: RunFailed, Plan: planPtr(st),
-			WorkDir: st.Meta.WorkDir, Error: msg, Summary: msg}, nil
-	}
+	_ = st.save()
 	st.mu.Unlock()
 	summary := fmt.Sprintf("review：保留 %d 条 / 丢弃 %d 条未验证", len(verified), dropped)
 	if finderFailed > 0 {

squad/runstate.go

@@ -96,11 +96,6 @@ func (st *RunState) persist() error {
 			return err
 		}
 	}
-	// steering.json 与其他 RunState 字段同权：SaveRun 后的 LoadRun 必须能读回
-	// 未消费的补充指令（与 SQLiteStore 的 steering 列同一契约）。
-	if err := writeJSON(filepath.Join(st.Meta.WorkDir, "steering.json"), st.Steering); err != nil {
-		return err
-	}
 	return writeJSON(filepath.Join(st.Meta.WorkDir, "runstate.json"), diskRunState{
 		Approved:              st.Approved,
 		RepairBudgetRemaining: st.RepairBudgetRemaining,

squad/runstore.go

@@ -5,7 +5,6 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
-	"sync"
 	"time"
 )
 
@@ -110,17 +109,6 @@ type StepSnapshot struct {
 	ID        string `json:"id"`
 	Status    string `json:"status"`
 	AttemptID string `json:"attempt_id,omitempty"`
-	// F05/U3b：trusted terminal record；用于 recovery 时仅在「后端显式
-	// 声明 success」时把步骤恢复成 completed（即便目标文件缺失也照恢复），
-	// files 仍然记入 deliverables 做诊断。SideEffects 报告重放安全性：
-	//   none     → 可重试
-	//   possible → 需 reconciliation（不禁 rerun；只允许显式 reconcile）
-	// Canceled/Aborted 为 true 时本步是取消/中止语义，不应被恢复成 completed。
-	Terminal string `json:"terminal,omitempty"`
-	// SideEffects 镜像 ExecResult.SideEffects；缺省 = possible（旧读取）。
-	SideEffects string `json:"side_effects,omitempty"`
-	Canceled    bool   `json:"canceled,omitempty"`
-	Aborted     bool   `json:"aborted,omitempty"`
 }
 
 // CheckpointMode 快照写盘档位（Q1 WP4 三档）。
@@ -141,10 +129,6 @@ type checkpointWriter struct {
 	store Store
 	ch    chan Checkpoint
 	done  chan error
-	// flushOnce/flushErr 让出口 flush 幂等：正常收口路径先 flush 一次（失败必须
-	// 先于任何完成事件可见），defer 再兜底 flush 时不会二次写、也不会重复报错。
-	flushOnce sync.Once
-	flushErr  error
 }
 
 func newCheckpointWriter(mode CheckpointMode, store Store) *checkpointWriter {
@@ -194,13 +178,7 @@ func (w *checkpointWriter) boundary(wsDir string, cp Checkpoint) error {
 }
 
 // flush 在推进退出前调用：async 落完并报告首个错误；on_exit 写最终快照。
-// 幂等：重复调用返回首次结果（async 的 chan 只关一次）。
 func (w *checkpointWriter) flush(wsDir string, cp Checkpoint) error {
-	w.flushOnce.Do(func() { w.flushErr = w.flushOnceImpl(wsDir, cp) })
-	return w.flushErr
-}
-
-func (w *checkpointWriter) flushOnceImpl(wsDir string, cp Checkpoint) error {
 	cp.wsDir = wsDir
 	switch w.mode {
 	case CheckpointAsync:
@@ -213,15 +191,6 @@ func (w *checkpointWriter) flushOnceImpl(wsDir string, cp Checkpoint) error {
 	}
 }
 
-// persistenceFailurePrefix 标记落盘失败。这类失败必须让运行以可区分的面目收场：
-// 业务动作（步骤执行、交付物写出）已发生，不回滚、不声称完成。
-const persistenceFailurePrefix = "persistence failure"
-
-// persistenceFailureText 组装落盘失败的可读说明（stage = 落盘环节）。
-func persistenceFailureText(stage string, err error) string {
-	return persistenceFailurePrefix + ": " + stage + ": " + err.Error()
-}
-
 // validateGraphContinuity 恢复连续性校验（Q1 WP4；WP1 的图版本裁定）。
 // 最新检查点存在且其图指纹与当前 Steps 重编译不一致时拒绝恢复——
 // 崩溃期间图形态被改过，继续推进会把已完成的结果接到错误的图上。

squad/step_retry.go

@@ -110,9 +110,6 @@ func AssessStepFailure(sig StepFailureSignals) FailureAssessment {
 	lower := strings.ToLower(reason)
 	trace := ScanTraceText(sig.ExtraText + "\n" + ReadWorkDirTraces(sig.WorkDir))
 	missing := sig.TargetMissing || strings.Contains(lower, "missing target deliverable")
-	// F05/KTD4：explicit SideEffects=none 报告 = provably-not-dispatched。
-	// 缺则按 possible 对待（与旧语义一致）。这是 caller 自己传达的可重试信令。
-	provablyNotDispatched := sig.Result != nil && sig.Result.SideEffects == ExecSideEffectsNone
 	if sig.TargetExists {
 		return FailureAssessment{Class: FailurePermanent, Reason: "target-already-present: do not overwrite"}
 	}
@@ -138,15 +135,6 @@ func AssessStepFailure(sig StepFailureSignals) FailureAssessment {
 	if missing && trace.Settled && !provider {
 		return FailureAssessment{Class: FailureTransient, Reason: "settled-no-deliverable: " + compactReason(reason)}
 	}
-	// F05/U3b：SideEffects=none 失败原则上可重试（后端已声明未发生外部动作）；
-	// 跨越该含义补一条有界重试通路，并覆盖纯获取后失败但文件已拼起的边界。
-	if provablyNotDispatched && !provider {
-		kind := "provably-not-dispatched"
-		if missing {
-			kind = "provably-not-dispatched-missing-target"
-		}
-		return FailureAssessment{Class: FailureTransient, Reason: kind + ": " + compactReason(reason)}
-	}
 	if missing {
 		return FailureAssessment{Class: FailurePermanent, Reason: "permanent-missing: " + compactReason(reason)}
 	}

squad/workflow.go

@@ -3,7 +3,6 @@ package squad
 import (
 	"bytes"
 	"context"
-	"encoding/json"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -42,41 +41,16 @@ func (d *flowDriver) advance(ctx context.Context, st *RunState) (res *RunResult,
 	cpw := newCheckpointWriter(CheckpointMode(st.Meta.Metadata["checkpoint_mode"]), st.storeOr())
 	// 推进退出前 flush：async 落完、on_exit 写最终快照（Q1 WP4 三档的出口语义）。
 	var lastGraph *graph
-	// exitSnapshot 组装出口快照（尚未编译图时现编译）。
-	exitSnapshot := func() (Checkpoint, bool) {
-		if lastGraph == nil {
-			g, gerr := compileGraph(st.Plan.Workflow.Steps)
-			if gerr != nil {
-				return Checkpoint{}, false
-			}
-			lastGraph = g
-		}
-		return snapshotOf(superNo, lastGraph, st.Plan.Workflow.Steps), true
-	}
-	// 兜底出口：任何非正常退出路径（挂起、图编译失败、infra 错误）也在此 flush。
-	// flush 幂等——收口路径已提前 flush 过，不会二次写。
 	defer func() {
-		if cp, ok := exitSnapshot(); ok {
-			if ferr := cpw.flush(st.Meta.WorkDir, cp); ferr != nil {
-				// F11c：出口快照落盘失败不得静默——翻成可区分的失败。
-				res, gate, err = d.persistenceFailure(st, res, gate, err, "exit checkpoint flush", ferr)
+		if lastGraph == nil {
+			if g, gerr := compileGraph(st.Plan.Workflow.Steps); gerr == nil {
+				lastGraph = g
 			}
 		}
+		if lastGraph != nil {
+			_ = cpw.flush(st.Meta.WorkDir, snapshotOf(superNo, lastGraph, st.Plan.Workflow.Steps))
+		}
 	}()
-	// flushExitBeforeFinalize 在收口前落出口快照（async/on_exit 的最终落盘），
-	// 失败立即交出可区分的失败：落盘失败必须先于任何完成事件/RunCompleted 返回。
-	// 返回的 ok=false 表示快照已落盘（或本次无需落盘），调用方继续正常收口。
-	flushExitBeforeFinalize := func() (*RunResult, *HitlGate, error, bool) {
-		cp, ok := exitSnapshot()
-		if !ok {
-			return nil, nil, nil, false
-		}
-		if ferr := cpw.flush(st.Meta.WorkDir, cp); ferr != nil {
-			out, g2, e2 := d.persistenceFailure(st, nil, nil, nil, "exit checkpoint flush", ferr)
-			return out, g2, e2, true
-		}
-		return nil, nil, nil, false
-	}
 	for {
 		if !runDeadline.IsZero() && time.Now().After(runDeadline) {
 			d.emit(st, Event{Type: EvPhase, Text: "运行截止时刻已到，未完步骤按 skipped 收口"})
@@ -89,9 +63,6 @@ func (d *flowDriver) advance(ctx context.Context, st *RunState) (res *RunResult,
 			}
 			_ = st.save()
 			st.mu.Unlock()
-			if out, g2, e2, failed := flushExitBeforeFinalize(); failed {
-				return out, g2, e2
-			}
 			return d.finalize(ctx, st)
 		}
 		g, gerr := compileGraph(st.Plan.Workflow.Steps)
@@ -180,39 +151,21 @@ func (d *flowDriver) advance(ctx context.Context, st *RunState) (res *RunResult,
 			}
 		}
 
-		// superstep 边界快照（sync 立即 / async 入队；on_exit 由出口 flush 写）。
+		// superstep 边界快照（sync 立即 / async 入队；on_exit 由 defer 的 flush 写）。
 		lastGraph = g
 		if err := cpw.boundary(st.Meta.WorkDir, snapshotOf(superNo, g, st.Plan.Workflow.Steps)); err != nil {
-			// F11c：边界快照写盘失败已经是可见错误，补上落盘失败标记以便识别。
-			return nil, nil, fmt.Errorf("%s: %w", persistenceFailurePrefix, err)
+			return nil, nil, err
 		}
 	}
 
-	// 全部步骤已收口：先落出口快照（async/on_exit 的最终落盘），失败必须先于收口
-	// 可见——不得先宣布完成、再报落盘失败。
-	if out, g2, e2, failed := flushExitBeforeFinalize(); failed {
-		return out, g2, e2
-	}
-
 	return d.finalize(ctx, st)
 }
 
 // snapshotOf 组装一个 superstep 边界快照：序号、图结构指纹与全量步骤状态。
-// terminal side-effect evidence 由 recordStepTerminal 在步骤进入 completed 时写入
-// step_terminals.json；本处把已记录的 terminal 数据并入 StepSnapshot，保证
-// checkpoint 也带同样证据（F05：信任落盘记录）。
 func snapshotOf(superNo int, g *graph, steps []StepSpec) Checkpoint {
 	cp := Checkpoint{Superstep: superNo, At: time.Now(), GraphHash: g.hash()}
-	// checkpoint 路径拿不到当前 RunState 引用（cpw 在 defer flush 时调用过），
-	// 同时也仅有顺序式调用传递上下文不便；所以这里走「按 ID 检查同索引」对齐
-	// 替代——checkpoint 产生时 step.Status 已为终态（completed），因此 record 同步
-	// 已写入；StepSnapshot.Terminal 就用 step.Status 推导（F05 唯一信任源）。
 	for _, s := range steps {
-		snap := StepSnapshot{ID: s.ID, Status: s.Status, AttemptID: s.AttemptID}
-		if s.Status == "completed" {
-			snap.Terminal = "completed"
-		}
-		cp.StepStates = append(cp.StepStates, snap)
+		cp.StepStates = append(cp.StepStates, StepSnapshot{ID: s.ID, Status: s.Status, AttemptID: s.AttemptID})
 	}
 	return cp
 }
@@ -406,9 +359,6 @@ func (d *flowDriver) runOneStep(ctx context.Context, st *RunState, idx int) (*Ru
 	}
 	var result *ExecResult
 	var err error
-	// F05/U3b（KTD4）：ExecutionID 在一次逻辑执行内稳定。重试层共享同一 ID
-	// + 同一 Deadline，便于下游做幂等键 + 收据对齐。第一层 attempt 用稳定 ID；
-	// 重试层沿用 step.AttemptID（首次进入时已生成、保持不变）。
 	for attempt := 1; attempt <= MaxTransientStepInvocations; attempt++ {
 		reqPrompt := prompt
 		if attempt > 1 {
@@ -429,15 +379,14 @@ func (d *flowDriver) runOneStep(ctx context.Context, st *RunState, idx int) (*Ru
 			ToolProfile:  orDefault(role.ToolProfile, ToolProfileWorkspace),
 			OnEvent: func(ev ProcessEvent) {
 				d.emit(st, Event{Type: EvDetail, StepID: step.ID, Role: step.Role, Text: ev.Text,
-					Data: processEventData(ev)})
+					Data: marshalJSON(map[string]string{"process_type": ev.Type, "tool": ev.Tool, "detail": ev.Detail})})
 			},
 			// RoleSpec v2 贯通：工具白名单 / 技能 / 标识（原为死字段或缺失）。
 			ToolAllowlist: role.Tools,
 			Skills:        firstNonEmptySlice(step.Skills, role.Skills),
-			// F05：ExecutionID 复用 step.AttemptID（首层 attempt 时已生成）——重试层不再换新 ID。
-			SessionScope: st.Meta.ID + "/" + step.ID,
-			ExecutionID:  step.AttemptID,
-			Budget:       stepBudget,
+			SessionScope:  st.Meta.ID + "/" + step.ID,
+			ExecutionID:   step.AttemptID,
+			Budget:        stepBudget,
 		}
 		if step.IntentID != "" {
 			// 系统功能步：不经智能体，由 IntentRunner 直接执行；缺席即 fail-closed。
@@ -457,59 +406,18 @@ func (d *flowDriver) runOneStep(ctx context.Context, st *RunState, idx int) (*Ru
 		} else {
 			result, err = backend.Execute(ctx, req)
 		}
-		// F05/KTD4：先验证执行结果本身，不让「文件已存在」翻成 Success。
-		// (1) (nil,nil) 视作稳定失败，不允许 panic 也不允许误判成功；
-		// (2) Success=true 仅在 result 显式声明时才成立；
-		// (3) 已取消/中止的结果不进入重试通路（不是业务失败，且重试会重复已发生的取消语义）。
-		if err != nil && result == nil {
-			// (nil,nil) 在 Go 是不可能态（要么 err、要么 result）；保险起见仍按稳定失败。
-			step.Status = "failed"
-			step.Error = "executor returned no result and no error (stable failure)"
-			break
-		}
-		if result == nil {
-			step.Status = "failed"
-			step.Error = "executor returned nil result"
-			break
-		}
-		if result.IsCanceled() {
-			// F05/KTD4：取消不是业务失败，不重试（按 attempt==MaxTransientStepInvocations 跳出）。
-			step.Status = "failed"
-			if result.ErrorReason != "" {
-				step.Error = "executor canceled: " + result.ErrorReason
-			} else {
-				step.Error = "executor canceled"
-			}
-			break
-		}
 		if err == nil && result.Success {
-			// 交付物物化：后端返回 patches 但未写盘时补写。F05：仅当 result.Success
-			// 已验证且文件未准备时才需要物化；只有 patches 内容才能落盘。
-			if len(result.Patches) > 0 {
-				_, _ = MaterializeTargets(subDir, targets, result.Patches)
-			}
 			break
 		}
-		// 交付物物化：后端返回 patches 但未写盘时补写。F05：仅当 result.Success 已
-		// 被后端显式声明时物化才允许 break；旧逻辑会让
-		// `err==nil && len(Patches)>0 && materialize_ok` 翻成 Success=true —— 修掉。
+		// 交付物物化：后端返回 patches 但未写盘时补写。
 		if err == nil && len(result.Patches) > 0 {
 			if _, merr := MaterializeTargets(subDir, targets, result.Patches); merr == nil {
-				if result.Success && allTargetsPresent(subDir, targets) {
+				if allTargetsPresent(subDir, targets) {
+					result.Success = true
 					break
 				}
 			}
 		}
-		// F05/KTD4：副作用可能存在的失败禁止自动重试（重试会重复外部动作）。
-		// 仅 SideEffects=none 或纯读取后的失败保持有界重试。
-		if !result.IsRetryableSideEffectReports() && err != nil {
-			step.Status = "failed"
-			step.Error = orDefault(result.ErrorReason, err.Error())
-			if result.ErrorReason != "" {
-				step.Error = result.ErrorReason
-			}
-			break
-		}
 		assessment := AssessStepFailure(StepFailureSignals{
 			Err: err, Result: result, WorkDir: subDir,
 			ExtraText:     fmt.Sprintf("step=%s backend=%s", step.ID, backendName),
@@ -555,9 +463,6 @@ func (d *flowDriver) runOneStep(ctx context.Context, st *RunState, idx int) (*Ru
 
 	step.Status = "completed"
 	appendLog(subDir, fmt.Sprintf("[%s] backend=%s duration=%dms targets=%v\n", time.Now().Format(time.RFC3339), backendName, step.DurationMs, targets))
-	// F05/U3b：记录 trusted terminal state（用于 recoverInterrupted 判定）。
-	// 由「正在收口的 completed」流写入；recovery 期间不会回填（防伪）。
-	d.recordStepTerminal(st, step, result)
 
 	// 状态回填 + 黑板合并 + 落盘同一临界区：并行步骤的共享状态变更串行化，
 	// 防止 marshal 与写回竞争、临时文件互相踩踏把保存失败误判为步骤失败。
@@ -990,22 +895,10 @@ func (d *flowDriver) finalize(ctx context.Context, st *RunState) (*RunResult, *H
 	st.setWorkflow(wf)
 	st.Meta.Status = string(status)
 	st.Meta.CompletedAt = &now
-	// F11c：终态落盘先于完成事件。写不进去就不能宣布完成——
-	// 业务动作已发生，不回滚，但运行以可区分的失败面世。
-	if serr := st.save(); serr != nil {
-		msg := persistenceFailureText("save terminal run state", serr)
-		return &RunResult{RunID: st.Meta.ID, Status: RunFailed, Plan: planPtr(st),
-			Quality: report, WorkDir: st.Meta.WorkDir, Usage: st.usageSnapshot(),
-			Error: msg, Summary: msg}, nil, nil
-	}
+	_ = st.save()
 
 	manifest := buildManifest(st, deliverable, *report)
-	if merr := writeJSON(filepath.Join(st.Meta.WorkDir, "manifest.json"), manifest); merr != nil {
-		msg := persistenceFailureText("write manifest", merr)
-		return &RunResult{RunID: st.Meta.ID, Status: RunFailed, Plan: planPtr(st),
-			Quality: report, WorkDir: st.Meta.WorkDir, Usage: st.usageSnapshot(),
-			Error: msg, Summary: msg}, nil, nil
-	}
+	_ = writeJSON(filepath.Join(st.Meta.WorkDir, "manifest.json"), manifest)
 
 	d.emit(st, Event{Type: EvRunCompleted, Text: string(status),
 		Data: marshalJSON(map[string]any{"deliverable": deliverable, "score": report.Score})})
@@ -1125,41 +1018,6 @@ func (d *flowDriver) emitHuman(st *RunState, req *HumanRequest) {
 	d.emit(st, Event{Type: EvHumanInputRequired, Text: req.Question, Data: marshalJSON(req)})
 }
 
-// persistenceFailure 把一次落盘失败折算成可见的失败（F11c）：
-//   - 已有 infra 错误 → 附加到错误里返回，不吞掉原错误；
-//   - 挂起中 → 挂起语义不变（运行未收口），但结果与事件流如实标注；
-//   - 其余（收口完成/无结果）→ 结果翻成 RunFailed，绝不再以 completed 面世。
-//
-// 业务动作（步骤执行、交付物写出）已发生，不回滚，也不声称干净完成。
-func (d *flowDriver) persistenceFailure(st *RunState, res *RunResult, gate *HitlGate, prev error, stage string, cause error) (*RunResult, *HitlGate, error) {
-	msg := persistenceFailureText(stage, cause)
-	if prev != nil {
-		return res, gate, fmt.Errorf("%w（同时发生 %s）", prev, msg)
-	}
-	if res == nil {
-		res = &RunResult{RunID: st.Meta.ID, Status: RunFailed, Plan: planPtr(st),
-			WorkDir: st.Meta.WorkDir, Summary: msg}
-	}
-	if !strings.Contains(res.Error, msg) {
-		// 同一次失败会经出口 flush 与兜底 flush 各报一次，只记一遍。
-		if res.Error != "" {
-			msg = res.Error + "；" + msg
-		}
-		res.Error = msg
-	}
-	if gate != nil {
-		// 挂起路径不发终态事件（finalizeEvent 只处理 completed/degraded/failed），
-		// 所以在这里补一条可观察的标注事件。
-		d.emit(st, Event{Type: EvPhase, Text: msg})
-		return res, gate, nil
-	}
-	switch res.Status {
-	case RunCompleted, RunDegraded:
-		res.Status = RunFailed
-	}
-	return res, nil, nil
-}
-
 func summarize(st *RunState, report QualityReport, status RunStatus) string {
 	var b strings.Builder
 	fmt.Fprintf(&b, "运行 %s：%s。步骤 %d 个（完成 %d/跳过 %d/失败 %d）。质检得分 %.0f。",
@@ -1181,84 +1039,6 @@ func countStatus(st *RunState, status string) int {
 	return n
 }
 
-// recordStepTerminal 记一条「trusted terminal record」：在某步执行完成
-// （仅 success 路径）时落地 step.Terminal + SideEffects + canceled/aborted。
-// reconcile/path：如果 recovery 期间没有该记录，则按文件存在与否分流
-// （见 engine.recoverInterrupted）。
-//
-// F05：record 与 read 路径严格同包；测试通过 squad/executor_result_contract_test.go
-// 覆盖正向与逆向条件。
-func (d *flowDriver) recordStepTerminal(st *RunState, step StepSpec, result *ExecResult) {
-	if st == nil || st.Meta.WorkDir == "" {
-		return
-	}
-	if step.ID == "" {
-		return
-	}
-	t := stepTerminalRecord{Terminal: "completed", At: time.Now(), StepID: step.ID, AttemptID: step.AttemptID}
-	if result != nil {
-		t.SideEffects = string(result.SideEffects)
-		t.Canceled = result.Canceled
-		t.Aborted = result.Aborted
-	}
-	st.mu.Lock()
-	defer st.mu.Unlock()
-	terminals := readStepTerminalsLocked(st)
-	terminals[step.ID] = t
-	data, err := json.Marshal(terminals)
-	if err != nil {
-		return
-	}
-	_ = os.WriteFile(filepath.Join(st.Meta.WorkDir, "step_terminals.json"), data, 0o644)
-}
-
-// readStepTerminalsLocked 读取已落盘的 step_terminals.json（调用方须持锁或私有化）。
-// 失败（含文件不存在）返回空 map。
-func readStepTerminalsLocked(st *RunState) map[string]stepTerminalRecord {
-	out := map[string]stepTerminalRecord{}
-	if st == nil || st.Meta.WorkDir == "" {
-		return out
-	}
-	data, err := os.ReadFile(filepath.Join(st.Meta.WorkDir, "step_terminals.json"))
-	if err != nil || len(data) == 0 {
-		return out
-	}
-	var stored map[string]stepTerminalRecord
-	if err := json.Unmarshal(data, &stored); err != nil {
-		return out
-	}
-	for k, v := range stored {
-		out[k] = v
-	}
-	return out
-}
-
-// loadStepTerminalsForWorkDir 公开暴露的 record 读取（recovery 使用）。
-func loadStepTerminalsForWorkDir(wsDir string) map[string]stepTerminalRecord {
-	if wsDir == "" {
-		return map[string]stepTerminalRecord{}
-	}
-	data, err := os.ReadFile(filepath.Join(wsDir, "step_terminals.json"))
-	if err != nil || len(data) == 0 {
-		return map[string]stepTerminalRecord{}
-	}
-	var stored map[string]stepTerminalRecord
-	if err := json.Unmarshal(data, &stored); err != nil {
-		return map[string]stepTerminalRecord{}
-	}
-	return stored
-}
-
-type stepTerminalRecord struct {
-	StepID      string    `json:"step_id"`
-	AttemptID   string    `json:"attempt_id"`
-	Terminal    string    `json:"terminal"` // completed (唯一允许值；failure 走 step.Status)
-	SideEffects string    `json:"side_effects,omitempty"`
-	Canceled    bool      `json:"canceled,omitempty"`
-	Aborted     bool      `json:"aborted,omitempty"`
-	At          time.Time `json:"at"`
-}
-
 func buildManifest(st *RunState, deliverable string, report QualityReport) Manifest {
 	files := map[string]string{}
 	_ = filepath.Walk(st.Meta.WorkDir, func(path string, info os.FileInfo, err error) error {