feat(workflow): add workflow engine, /workflows panel, /ultracode skill

将 feat/sdk-backend 分支中 workflow 相关的 20 个 commit 压缩为单 commit：

- 工作流引擎核心：phase / agent / parallel / pipeline 编排原语（packages/workflow-engine/）
- /workflows 面板：三区焦点布局（顶部 run tabs + 左侧 phase 侧栏 + 右侧 agent 列表）
- /ultracode skill：多 agent workflow 编排入口
- 进度存储 / journal / notification 系统
- WorkflowService 生命周期管理 + SentryErrorBoundary
- 脚本沙箱：禁用 dynamic import()、JSON args 防御性归一化
- journal 与 named-workflow 路径统一在 projectRoot
- 错误处理：parallel/pipeline hooks 错误日志、failure routing、semaphore abort
- workflow 工具升级为 core 工具 + PascalCase 命名

Co-Authored-By: glm-5.1 <zai-org@claude-code-best.win>

src/workflow/backends/claudeCodeBackend.ts

@@ -0,0 +1,158 @@
+// 深度集成后端：从活会话解析 agent/model/tools，委托核心 runAgent。
+// 实现 AgentAdapter 接口，由 registry（U5）注册并路由。
+import {
+  type AgentAdapter,
+  type AgentAdapterContext,
+  type AgentRunParams,
+  type AgentRunResult,
+} from '@claude-code-best/workflow-engine'
+import { assembleToolPool } from '../../tools.js'
+import { finalizeAgentTool } from '@claude-code-best/builtin-tools/tools/AgentTool/agentToolUtils.js'
+import { runAgent } from '@claude-code-best/builtin-tools/tools/AgentTool/runAgent.js'
+import {
+  isBuiltInAgent,
+  type AgentDefinition,
+  type BuiltInAgentDefinition,
+} from '@claude-code-best/builtin-tools/tools/AgentTool/loadAgentsDir.js'
+import { createUserMessage, extractTextContent } from '../../utils/messages.js'
+import { createAgentId } from '../../utils/uuid.js'
+import { logForDebugging } from '../../utils/debug.js'
+import { logEvent } from '../../services/analytics/index.js'
+import type { ModelAlias } from '../../utils/model/aliases.js'
+import type { Message } from '../../types/message.js'
+import type { ToolUseContext } from '../../Tool.js'
+import { readHostBundle } from '../hostHandle.js'
+
+/** workflow 子 agent 的兜底定义（agentType 未命中真实注册表时用）。 */
+export const WORKFLOW_AGENT: BuiltInAgentDefinition = {
+  agentType: 'workflow-worker',
+  whenToUse: 'workflow 脚本内 agent() 钩子派发的子任务',
+  tools: ['*'],
+  source: 'built-in',
+  baseDir: 'built-in',
+  getSystemPrompt: () =>
+    'You are a workflow sub-agent. Complete the task concisely; your final text is the return value relayed to the workflow.',
+}
+
+/** agentType → 真实 agent 注册表（activeAgents 命中即用，否则兜底）。已导出便于单测。 */
+export function resolveAgentDefinition(
+  agentType: string | undefined,
+  toolUseContext: ToolUseContext,
+): AgentDefinition {
+  if (!agentType) return WORKFLOW_AGENT
+  const found = toolUseContext.options.agentDefinitions.activeAgents.find(
+    a => a.agentType === agentType,
+  )
+  return found ?? WORKFLOW_AGENT
+}
+
+/** model 别名 → 当前 provider 实际 model id。v1 直传（保留映射扩展点）。已导出便于单测。 */
+export function mapWorkflowModel(
+  model: string | undefined,
+): string | undefined {
+  return model
+}
+
+/** 从 agent 最终消息中提取 StructuredOutput 产出的 JSON 对象；失败返回 null。已导出便于单测。 */
+export function extractStructuredOutput(
+  content: Array<{ type: string; text?: string }>,
+): unknown | null {
+  for (const block of content) {
+    if (block.type === 'text' && block.text) {
+      const trimmed = block.text.trim()
+      const start = trimmed.indexOf('{')
+      const end = trimmed.lastIndexOf('}')
+      if (start >= 0 && end > start) {
+        try {
+          return JSON.parse(trimmed.slice(start, end + 1))
+        } catch {
+          // 继续尝试下一个文本块
+        }
+      }
+    }
+  }
+  return null
+}
+
+/** 深度集成后端：从活会话解析 agent/model/tools，委托核心 runAgent。 */
+export const claudeCodeBackend: AgentAdapter = {
+  id: 'claude-code',
+  capabilities: { structuredOutput: true, tools: true },
+
+  async run(
+    params: AgentRunParams,
+    ctx: AgentAdapterContext,
+  ): Promise<AgentRunResult> {
+    const { toolUseContext, canUseTool } = readHostBundle(ctx.host)
+    const appState = toolUseContext.getAppState()
+    const agentDef = resolveAgentDefinition(params.agentType, toolUseContext)
+    const model = mapWorkflowModel(params.model)
+    const agentId = createAgentId()
+
+    const workerPermissionContext = {
+      ...appState.toolPermissionContext,
+      mode: agentDef.permissionMode ?? 'acceptEdits',
+    }
+    const workerTools = assembleToolPool(
+      workerPermissionContext,
+      appState.mcp.tools,
+    )
+
+    // schema → 通过 prompt 追加 JSON Schema 指令（非交互模式 StructuredOutput 已启用）
+    const promptText = params.schema
+      ? `${params.prompt}\n\nYou MUST return your final answer by calling the StructuredOutput tool with a value matching this JSON Schema:\n${JSON.stringify(params.schema)}`
+      : params.prompt
+
+    const promptMessages = [createUserMessage({ content: promptText })]
+    const messages: Message[] = []
+    const startTime = Date.now()
+
+    try {
+      for await (const msg of runAgent({
+        agentDefinition: agentDef,
+        promptMessages,
+        toolUseContext,
+        canUseTool,
+        isAsync: true,
+        querySource: toolUseContext.options.querySource ?? 'workflow',
+        availableTools: workerTools,
+        override: { agentId },
+        // runAgent 的 model 是顶层 ModelAlias；workflow 的 model 是任意别名串，
+        // 类型上不兼容，运行时由 provider 层解析。双重断言透传（优于 as any/never）。
+        ...(model ? { model: model as unknown as ModelAlias } : {}),
+      })) {
+        messages.push(msg as Message)
+      }
+    } catch (e) {
+      logForDebugging(
+        `workflow sub-agent error (${agentDef.agentType}): ${(e as Error).message}`,
+      )
+      logEvent('tengu_workflow_agent', { ok: 0 })
+      return { kind: 'dead' }
+    }
+
+    const finalized = finalizeAgentTool(messages, agentId, {
+      prompt: params.prompt,
+      resolvedAgentModel: toolUseContext.options.mainLoopModel,
+      isBuiltInAgent: isBuiltInAgent(agentDef),
+      startTime,
+      agentType: agentDef.agentType,
+      isAsync: true,
+    })
+    const outputTokens =
+      finalized.usage?.output_tokens ?? finalized.totalTokens ?? 0
+    logEvent('tengu_workflow_agent', { ok: 1, outputTokens })
+
+    if (params.schema) {
+      const structured = extractStructuredOutput(finalized.content)
+      if (structured === null) return { kind: 'dead' }
+      return {
+        kind: 'ok',
+        output: structured as object,
+        usage: { outputTokens },
+      }
+    }
+    const text = extractTextContent(finalized.content, '\n')
+    return { kind: 'ok', output: text, usage: { outputTokens } }
+  },
+}