@dreb/代理核心

具有工具执行和事件流的状态代理 。 建造在 @dreb/ai。 。 。 .

安装

npm install @dreb/agent-core

快速启动

import { Agent } from "@dreb/agent-core";
import { getModel } from "@dreb/ai";

const agent = new Agent({
  initialState: {
    systemPrompt: "You are a helpful assistant.",
    model: getModel("anthropic", "claude-sonnet-4-20250514"),
  },
});

agent.subscribe((event) => {
  if (event.type === "message_update" && event.assistantMessageEvent.type === "text_delta") {
    // Stream just the new text chunk
    process.stdout.write(event.assistantMessageEvent.delta);
  }
});

await agent.prompt("Hello!");

核心概念

代理Message vs LLM 信件

代理与 AgentMessage,灵活类型,可以包括:

• 标准 LLM 消息 (user, (中文(简体) ), assistant, (中文(简体) ), toolResult(中文(简体) )
• 通过声明合并自定义应用特定消息类型

法学硕士只理解 user, (中文(简体) ), assistant,以及 toolResult编辑 convertToLlm 函数通过在每次LLM调用前过滤和转换消息来弥补这一空白.

信件流动

settled AgentContext → beforeLlmCall() → transformContext() → convertToLlm() → LLM
                         (optional)          (optional)         (required)

1. 在 LlmCall 之前: 检查已结算的用户/工具-结果上下文,并在请求前在原子上替换消息和/或活动模式
2. 转换语句: prune 旧信件或输入请求的外部上下文
3. 转换为Llm: 过滤出只包含 UI 的信件,并将自定义类型转换为 LLM 格式

beforeLlmCall 运行后 shouldContinue 允许申请及申请前 turn_start返回 { messages, model } 更新活动循环和 Agent.state; 返回时 undefined 保存他们。 钩子收到快照,所以没有回取替换的突变被忽略. 与 Agent,被抛出的错误会以可见的助手错误终止运行,而不是应用部分替换;低级循环调用器必须自表面出故障并返回 undefined 而不是拒绝.

事件流程

代理为 UI 更新发布事件 。 了解事件序列有助于构建响应界面.

提示( ) 事件顺序

当你打电话 prompt("Hello")数字 :

prompt("Hello")
├─ agent_start
├─ turn_start
├─ message_start   { message: userMessage }      // Your prompt
├─ message_end     { message: userMessage }
├─ message_start   { message: assistantMessage } // LLM starts responding
├─ message_update  { message: partial... }       // Streaming chunks
├─ message_update  { message: partial... }
├─ message_end     { message: assistantMessage } // Complete response
├─ turn_end        { message, toolResults: [] }
└─ agent_end       { messages: [...] }

用工具调用

如果助手调用工具,循环继续:

prompt("Read config.json")
├─ agent_start
├─ turn_start
├─ message_start/end  { userMessage }
├─ message_start      { assistantMessage with toolCall }
├─ message_update...
├─ message_end        { assistantMessage }
├─ tool_execution_start  { toolCallId, toolName, args }
├─ tool_execution_update { partialResult }           // If tool streams
├─ tool_execution_end    { toolCallId, result }
├─ message_start/end  { toolResultMessage }
├─ turn_end           { message, toolResults: [toolResult] }
│
├─ turn_start                                        // Next turn
├─ message_start      { assistantMessage }           // LLM responds to tool result
├─ message_update...
├─ message_end
├─ turn_end
└─ agent_end

工具执行模式可配置 :

• parallel (默认):飞行前工具按顺序调用,同时执行允许的工具,发送最终 tool_execution_end 财务报告和审定财务报表 toolResult 消息以助理源顺序排列
• sequential: 逐一执行工具调用, 匹配历史行为

这个 beforeToolCall 钩子在后面运行 tool_execution_start 并验证论证。 它可以阻止执行。 这个 afterToolCall 工具执行完成后及执行前的勾钩运行 tool_execution_end 发布最后的工具结果信息事件.

当你使用 Agent 班级,助理 message_end 在工具开始飞行前,加工被视为障碍。 这意味着 beforeToolCall 查看已包含请求工具调用助理消息的代理状态 .

持续( ) 事件顺序

continue() 从现有上下文恢复,而不添加新信件。 错误发生后用它进行重试 .

// After an error, retry from current state
await agent.continue();

上下文中的最后信息必须是 user 或者说 toolResult (无) assistant) (中文(简体) ).

事件类型

agent_start 将供应商/模型身份和有效思维级别发送给供应商,用于该运行:

{ type: "agent_start", model: { provider: "anthropic", id: "claude-sonnet-4-6" }, thinkingLevel: "high" }

代理选项

const agent = new Agent({
  // Initial state
  initialState: {
    systemPrompt: string,
    model: Model<any>,
    thinkingLevel: "off" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max",
    tools: AgentTool<any>[],
    messages: AgentMessage[],
  },

  // Prepare settled context immediately before each LLM request. Both fields
  // are optional; replacements apply to the active loop and Agent.state.
  beforeLlmCall: async (context, signal) => ({
    messages: compactedMessages,
    model: resizedContextModel,
  }),

  // Convert AgentMessage[] to LLM Message[] (required for custom message types)
  convertToLlm: (messages) => messages.filter(...),

  // Transform context before convertToLlm (for pruning, compaction)
  transformContext: async (messages, signal) => pruneOldMessages(messages),

  // Steering mode: "one-at-a-time" (default) or "all"
  steeringMode: "one-at-a-time",

  // Follow-up mode: "one-at-a-time" (default) or "all"
  followUpMode: "one-at-a-time",

  // Custom stream function (for proxy backends)
  streamFn: streamProxy,

  // Session ID for provider caching
  sessionId: "session-123",

  // Dynamic API key resolution (for expiring OAuth tokens)
  getApiKey: async (provider) => refreshToken(),

  // Tool execution mode: "parallel" (default) or "sequential"
  toolExecution: "parallel",

  // Preflight each tool call after args are validated. Can block execution.
  beforeToolCall: async ({ toolCall, args, context }) => {
    if (toolCall.name === "bash") {
      return { block: true, reason: "bash is disabled" };
    }
  },

  // Postprocess each tool result before final tool events are emitted.
  afterToolCall: async ({ toolCall, result, isError, context }) => {
    if (!isError) {
      return { details: { ...result.details, audited: true } };
    }
  },

  // Forward non-fatal provider warnings (e.g. malformed SSE/JSON) to the caller
  onWarning: (code, message) => console.warn(`[${code}] ${message}`),

  // Stream drop retry: number of retries when a provider stream ends without
  // its terminal event (default: 3). Set to 0 to disable. Messages with a
  // `refusal` (provider safety refusal) are never retried.
  streamRetries: 3,

  // Base delay for exponential backoff between stream retries in ms (default: 1000)
  streamRetryBaseDelayMs: 1000,

  // Length retry: number of retries when a turn ends with stopReason "length"
  // (model exhausted its configured output limit). Each retry uses that same limit.
  // Separate from streamRetries (default: 2). Set to 0 to disable.
  lengthRetries: 2,

  // Custom thinking budgets for token-based providers
  thinkingBudgets: {
    minimal: 128,
    low: 512,
    medium: 1024,
    high: 2048,
  },

  // Thinking display for adaptive-thinking Anthropic/Bedrock Claude models.
  // "summarized" returns thinking text; "omitted" returns only an encrypted
  // signature (lower latency). Ignored by other models. Opus 4.7+ default to
  // "omitted" at the API, so pass "summarized" to keep thinking visible.
  thinkingDisplay: "summarized",
});

代理国

interface AgentState {
  systemPrompt: string;
  model: Model<any>;
  thinkingLevel: ThinkingLevel;
  tools: AgentTool<any>[];
  messages: AgentMessage[];
  isStreaming: boolean;
  streamMessage: AgentMessage | null;  // Current partial during streaming
  pendingToolCalls: Set<string>;
  error?: string;
}

通过 agent.state在流出时, streamMessage 包含部分助理消息.

方法

提示

// Text prompt
await agent.prompt("Hello");

// With images
await agent.prompt("What's in this image?", [
  { type: "image", data: base64Data, mimeType: "image/jpeg" }
]);

// AgentMessage directly
await agent.prompt({ role: "user", content: "Hello", timestamp: Date.now() });

// Continue from current context (last message must be user or toolResult)
await agent.continue();

州管理层

agent.setSystemPrompt("New prompt");
agent.setModel(getModel("openai", "gpt-4o"));
agent.setThinkingLevel("medium");
agent.setTools([myTool]);
agent.setToolExecution("sequential");
agent.setBeforeLlmCall(async (context, signal) => ({ messages: context.messages }));
agent.setBeforeToolCall(async ({ toolCall }) => undefined);
agent.setAfterToolCall(async ({ toolCall, result }) => undefined);
agent.replaceMessages(newMessages);
agent.appendMessage(message);
agent.clearMessages();
agent.reset();  // Clear everything

会议和思考预算

agent.sessionId = "session-123";

agent.thinkingBudgets = {
  minimal: 128,
  low: 512,
  medium: 1024,
  high: 2048,
};

控制权

agent.abort();           // Cancel current operation
await agent.waitForIdle(); // Wait for completion

活动

const unsubscribe = agent.subscribe((event) => {
  console.log(event.type);
});
unsubscribe();

指导和后续行动

引导消息允许您在工具运行时中断代理 。 后续消息允许您排队工作, 否则代理将停止 .

agent.setSteeringMode("one-at-a-time");
agent.setFollowUpMode("one-at-a-time");

// While agent is running tools
agent.steer({
  role: "user",
  content: "Stop! Do this instead.",
  timestamp: Date.now(),
});

// After the agent finishes its current work
agent.followUp({
  role: "user",
  content: "Also summarize the result.",
  timestamp: Date.now(),
});

const steeringMode = agent.getSteeringMode();
const followUpMode = agent.getFollowUpMode();

agent.clearSteeringQueue();
agent.clearFollowUpQueue();
agent.clearAllQueues();

使用清除SteeringQueee, 清除后继Queee, 或清除所有已排队的消息 .

当旋转完成后检测到方向信息时:

1. 当前助理消息中的所有工具呼叫都已完成
2. 输入引导消息
3. LLM 在下一个回合响应

只有在没有更多的工具呼叫和引导消息时,才会检查后续消息. 如果有排队,则进行注射并再转一圈.

自定义信件类型

扩展 AgentMessage 通过声明合并:

declare module "@dreb/agent-core" {
  interface CustomAgentMessages {
    notification: { role: "notification"; text: string; timestamp: number };
  }
}

// Now valid
const msg: AgentMessage = { role: "notification", text: "Info", timestamp: Date.now() };

处理自定义类型 convertToLlm数字 :

const agent = new Agent({
  convertToLlm: (messages) => messages.flatMap(m => {
    if (m.role === "notification") return []; // Filter out
    return [m];
  }),
});

工具

使用 AgentTool数字 :

import { Type } from "@sinclair/typebox";

const readFileTool: AgentTool = {
  name: "read_file",
  label: "Read File",  // For UI display
  description: "Read a file's contents",
  parameters: Type.Object({
    path: Type.String({ description: "File path" }),
  }),
  execute: async (toolCallId, params, signal, onUpdate) => {
    const content = await fs.readFile(params.path, "utf-8");

    // Optional: stream progress
    onUpdate?.({ content: [{ type: "text", text: "Reading..." }], details: {} });

    return {
      content: [{ type: "text", text: content }],
      details: { path: params.path, size: content.length },
    };
  },
};

agent.setTools([readFileTool]);

处理错误

丢出错误 当工具失败时。 不返回错误消息为内容 .

execute: async (toolCallId, params, signal, onUpdate) => {
  if (!fs.existsSync(params.path)) {
    throw new Error(`File not found: ${params.path}`);
  }
  // Return content only on success
  return { content: [{ type: "text", text: "..." }] };
}

丢出的错误被代理抓住,并作为工具错误报告到 LLM isError: true。 。 。 .

代理用户

对于通过后端代理的浏览器应用:

import { Agent, streamProxy } from "@dreb/agent-core";

const agent = new Agent({
  streamFn: (model, context, options) =>
    streamProxy(model, context, {
      ...options,
      authToken: "...",
      proxyUrl: "https://your-server.com",
    }),
});

代理服务器流 ProxyAssistantMessageEvent编号 AssistantMessageEvents 无带宽重 partial 页:1 内容 error 事件应当前进 errorMessage 而对于安全拒绝,信息是 refusal 对象,所以重建后的信件保留 refusal 也从未被作为短暂的失败而被重审过.

低级 API

对于没有代理类的直接控制:

import { agentLoop, agentLoopContinue } from "@dreb/agent-core";

const context: AgentContext = {
  systemPrompt: "You are helpful.",
  messages: [],
  tools: [],
};

const config: AgentLoopConfig = {
  model: getModel("openai", "gpt-4o"),
  beforeLlmCall: async (settledContext, signal) => ({ messages: settledContext.messages }),
  convertToLlm: (msgs) => msgs.filter(m => ["user", "assistant", "toolResult"].includes(m.role)),
  toolExecution: "parallel",
  beforeToolCall: async ({ toolCall, args, context }) => undefined,
  afterToolCall: async ({ toolCall, result, isError, context }) => undefined,
};

const userMessage = { role: "user", content: "Hello", timestamp: Date.now() };

for await (const event of agentLoop([userMessage], context, config)) {
  console.log(event.type);
}

// Continue from existing context
for await (const event of agentLoopContinue(context, config)) {
  console.log(event.type);
}

这些低层溪流是可观测的. 它们保存事件顺序, 但是它们不会等待您的 Async 事件处理在以后的制作阶段继续之前解决 . beforeLlmCall 在提出请求之前,将进行解剖。 如果您需要普通消息事件处理,在工具飞行前起到屏障的作用,请使用 Agent 类而不是生 agentLoop() 或者说 agentLoopContinue()。 。 。 .

许可证

麻省理工学院