 调用迁移到 supervisor agents 的 stream 与 generate 调用)
如何把 Mastra 的 .network() 调用迁移到 supervisor agents 的 stream 与 generate 调用【免费下载链接】mastraMastra is the modern TypeScript framework for AI-powered applications and agents.项目地址: https://gitcode.com/GitHub_Trending/ma/mastra如果你还在用Agent.network()agent network 的 routing agent来协调多个子 agent、workflow 和 tool现在应该把它迁移到 supervisor agents 模式把.network()调用替换为Agent.stream()流式或Agent.generate()非流式并改用标准流式 chunk 处理结果。Mastra 官方文档明确标注 agent networks 已废弃deprecated将在未来的版本中移除.network()代码在移除之前仍可运行但新开发已聚焦于 supervisor agents。适用前提项目使用mastra/core中的Agent且现有代码通过.network()做多 agent 协调supervisor agentssubagents 模式从mastra/core1.8.0开始提供迁移前需确认本地mastra/core版本不低于该版本现有的 agent 配置agents、workflows、tools、memory保持不变迁移只改变调用方式和结果处理。迁移前的现状检查确认代码属于 .network() 模式.network()的典型写法是给一个 routing agent 配置agents/workflows/tools和Memory.network()依赖 memory 存储任务历史并判断任务是否完成然后调用.network()并迭代自定义事件流const result await routingAgent.network(Research AI in education) for await (const chunk of result) { if (chunk.type network-execution-event-step-finish) { console.log(chunk.payload.result) } }如果代码中同时出现approveNetworkToolCall()、declineNetworkToolCall()或resumeNetwork()说明你还有工具审批和挂起恢复流程迁移时需要对齐新的流式 chunktool-call-approval等见 Supervisor Agents 文档 的 Tool approval propagation 一节本文主路径只覆盖常规调用与事件处理。第一步把 .network() 换成 .stream() 或 .generate()核心改动只有一处调用入口。流式场景用.stream()通过stream.textStream或stream.fullStream迭代标准 chunk非流式场景用.generate()直接读取result.text。Before.network()const result await routingAgent.network(Research AI in education) for await (const chunk of result) { if (chunk.type network-execution-event-step-finish) { console.log(chunk.payload.result) } }Aftersupervisor agent .stream()const stream await supervisorAgent.stream(Research AI in education, { maxSteps: 10, }) for await (const chunk of stream.textStream) { process.stdout.write(chunk) }maxSteps选项限制 supervisor 的迭代次数它取代了.network()中隐式的迭代上限。非流式用法同样传入该选项const result await supervisorAgent.generate(Research AI in education, { maxSteps: 10, }) console.log(result.text)两个方法的参数细节可分别参考 Agent.stream() Reference 和 Agent.generate() Reference。第二步把路由指令改写为明确的 supervisor 指令.network()的 routing agent 靠通用指令和子 agent 描述来决定调用什么supervisor agents 机制相同但文档建议指令更具体说明有哪些可用资源、什么情况下用哪个、如何协调、以及何时认为任务完成。Before.network()时代的 routing agentconst routingAgent new Agent({ id: routing-agent, instructions: You are a network of researchers and writers..., agents: { researchAgent, writingAgent }, memory: new Memory(), })Aftersupervisor agentconst supervisorAgent new Agent({ id: supervisor-agent, instructions: You coordinate research and writing tasks using specialized agents. Available resources: - researchAgent: Gathers factual data and sources (returns bullet points) - writingAgent: Transforms research into narrative content (returns full paragraphs) Delegation strategy: 1. For research requests: Delegate to researchAgent first 2. For writing requests: Delegate to writingAgent (provide research if available) 3. For complex requests: Delegate to researchAgent first, then writingAgent Success criteria: - All user questions are fully answered - Response is well-formatted and complete - If information is incomplete, continue iterating, agents: { researchAgent, writingAgent }, memory: new Memory(), })同时给每个子 agent 增加description字段说明用途、返回格式和使用时机——supervisor 依据这些描述决定委派给谁const researchAgent new Agent({ id: research-agent, description: Specializes in gathering factual information and data on any topic. Returns concise bullet-point summaries with key facts and sources. Does not write full articles or narrative content., }) const writingAgent new Agent({ id: writing-agent, description: Transforms research material into well-structured written content. Produces full paragraphs and complete articles. Best used after research has been gathered., })注意agent 实例本身不再依赖name: Routing Agent这类网络式配置迁移指南中 after 示例保留了agents与memory与你原有配置一致无需删除。第三步按映射表替换 .network() 事件处理如果你在处理特定的.network()事件类型把它们替换为 supervisor agent 的标准 stream chunk 类型|.network()事件 | Supervisor agent chunk | | - | - | |routing-agent-start|step-start| |routing-agent-end|step-finish| |agent-execution-start|step-start委派时 | |agent-execution-event-text-delta|text-delta| |agent-execution-event-finish|step-finish| |network-execution-event-step-finish|step-finishfinishReason: stop| |network-object|object-delta配合 structuredOutput | |network-object-result|object配合 structuredOutput |也就是说判断最终步骤结束的条件从chunk.type network-execution-event-step-finish变为检测step-finish且finishReason: stop的 chunk结构化输出场景下.network()的objectStream/stream.object对应object-delta/objectchunk仍需传入structuredOutput选项。可选补齐 .network() 没有的委派控制以下能力是 supervisor agents 新增的不属于最小迁移路径按需添加。它们可以配置在 agent 的defaultOptions里也可以按调用传入委派钩子onDelegationStart/onDelegationComplete——在委派前后拦截、修改或拒绝委派const stream await supervisorAgent.stream(Research AI in education, { maxSteps: 10, delegation: { onDelegationStart: async context { console.log(Delegating to: ${context.primitiveId}) if (context.primitiveId research-agent) { return { proceed: true, modifiedPrompt: ${context.prompt}\n\nFocus on 2024-2025 data., modifiedMaxSteps: 5, } } if (context.iteration 8) { return { proceed: false, rejectionReason: Max iterations reached. Synthesize current findings., } } return { proceed: true } }, }, })onDelegationComplete在委派结束后触发可调用context.bail()停止 supervisor 循环或返回{ feedback: ... }把反馈写入 supervisor 的 memoryconst stream await supervisorAgent.stream(Research AI in education, { maxSteps: 10, delegation: { onDelegationComplete: async context { if (context.error) { context.bail() // Stop further delegations return { feedback: Delegation to ${context.primitiveId} failed: ${context.error}. Try a different approach., } } }, }, })消息过滤messageFilter——默认子 agent 会收到完整会话上下文可用messageFilter过滤敏感内容或限制消息数量示例中剔除含confidential的消息并只保留最近 10 条const stream await supervisorAgent.stream(Research AI in education, { maxSteps: 10, delegation: { messageFilter: ({ messages, primitiveId, prompt }) { return messages .filter(msg { const content typeof msg.content string ? msg.content : JSON.stringify(msg.content) return !content.includes(confidential) }) .slice(-10) }, }, })迭代监控onIterationComplete——每次 supervisor 循环迭代后触发可用于记录进度、注入反馈或提前停止返回{ continue: false }const stream await supervisorAgent.stream(Research AI in education, { maxSteps: 10, onIterationComplete: async context { console.log(Iteration ${context.iteration}/${context.maxIterations}) if (!context.text.includes(recommendations)) { return { continue: true, feedback: Please include specific recommendations in your analysis., } } if (context.text.length 1000 context.finishReason stop) { return { continue: false } } return { continue: true } }, })任务完成评分isTaskComplete——用 scorer 自动校验任务是否完成校验失败时 supervisor 继续迭代失败反馈会进入会话上下文。示例使用createScorer检查输出是否同时包含analysis和recommendation并通过onComplete打印完成状态import { createScorer } from mastra/core/evals const taskCompleteScorer createScorer({ id: task-complete, name: Task Completeness, }).generateScore(async context { const text (context.run.output || ).toString() const hasAnalysis text.includes(analysis) const hasRecommendations text.includes(recommendation) return hasAnalysis hasRecommendations ? 1 : 0 }) const stream await supervisorAgent.stream(Research AI in education, { maxSteps: 10, isTaskComplete: { scorers: [taskCompleteScorer], strategy: all, onComplete: async result { console.log(Task complete:, result.complete) }, }, })验证迁移结果文档给出的成功表现按调用方式区分流式for await (const chunk of stream.textStream)持续输出文本 chunkstep-finish且finishReason: stop的 chunk 表示一次完整执行结束对应旧network-execution-event-step-finish非流式const result await supervisorAgent.generate(...)后读取result.text即得到最终回答若接入了isTaskCompleteonComplete回调中result.complete会给出任务完成判定。如果迁移后行为异常先对照上面第三步的事件映射表检查 chunk 类型是否写对——旧事件名如network-execution-event-step-finish在新 API 中不会再出现。限制与后续.network()已废弃且将随版本移除Agent networks 文档 写明 will be removed in a future major release现有代码在移除前可继续运行官方建议尽快迁移supervisor agents 需要mastra/core1.8.0 及以上版本.network()的approveNetworkToolCall()/resumeNetwork()等审批与恢复流程没有出现在迁移指南中遇到这类代码请结合 Supervisor Agents 文档 的 Tool approval propagation子 agent 的审批请求会以tool-call-approval等 chunk 出现在父 agent 流中处理更完整的委派选项request context、结果引用、后台任务、版本覆盖等见 Subagents 文档。参考文档Migration: .network() to supervisor agentsAgent networksSupervisor Agents (Subagents)Agent.stream() ReferenceAgent.generate() Reference【免费下载链接】mastraMastra is the modern TypeScript framework for AI-powered applications and agents.项目地址: https://gitcode.com/GitHub_Trending/ma/mastra创作声明：本文部分内容由AI辅助生成（AIGC），仅供参考
