open-agent-sdk 0.1.0-alpha.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Octane
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,284 @@
+# Open Agent SDK
+
+[![Build in Public](https://img.shields.io/badge/Build%20in%20Public-blue)](https://twitter.com/octane0411)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![TypeScript](https://img.shields.io/badge/TypeScript-5.0+-blue)](https://www.typescriptlang.org/)
+
+An open-source alternative to Claude Agent SDK — lightweight, customizable, and provider-agnostic.
+
+[中文文档](./README.zh.md)
+
+---
+
+## What is this?
+
+Open Agent SDK is a TypeScript framework for building AI agents. It provides a developer experience similar to Claude Agent SDK but with full transparency and no vendor lock-in.
+
+**Key features:**
+- **ReAct Loop** — Observation-thought-action cycle for autonomous agents
+- **Built-in Tools** — File operations (read/write/edit), shell execution, code search (glob/grep), web search
+- **Streaming Support** — Real-time response streaming with token usage tracking
+- **Multi-Provider** — Works with OpenAI and Google Gemini
+- **Session Management** — Persistent conversations with InMemory and File storage
+- **Permission System** — 4 permission modes (default/acceptEdits/bypassPermissions/plan)
+- **Hooks Framework** — Event-driven extensibility (9 hook events)
+- **Subagent System** — Delegate tasks to specialized agents
+- **Type Safety** — Full TypeScript support with strict type constraints
+- **Cancellation** — AbortController support for interrupting long-running operations
+
+## Installation
+
+```bash
+npm install open-agent-sdk
+```
+
+Or with specific package manager:
+
+```bash
+# npm
+npm install open-agent-sdk
+
+# yarn
+yarn add open-agent-sdk
+
+# pnpm
+pnpm add open-agent-sdk
+
+# bun
+bun add open-agent-sdk
+```
+
+**Requirements:**
+- Bun >= 1.0.0 (primary runtime)
+- Node.js >= 20 (with peer dependencies)
+- TypeScript >= 5.0
+
+## Quick Start
+
+### Basic Usage
+
+```typescript
+import { prompt } from 'open-agent-sdk';
+
+const result = await prompt("What files are in the current directory?", {
+  model: 'gpt-4o',
+  apiKey: process.env.OPENAI_API_KEY,
+});
+
+console.log(result.result);
+console.log(`Duration: ${result.duration_ms}ms`);
+console.log(`Tokens: ${result.usage.input_tokens} in / ${result.usage.output_tokens} out`);
+```
+
+### Using Gemini
+
+```typescript
+const result = await prompt("Explain quantum computing", {
+  model: 'gemini-2.0-flash',
+  provider: 'google',
+  apiKey: process.env.GEMINI_API_KEY,
+});
+```
+
+### Session-Based Conversations
+
+```typescript
+import { createSession } from 'open-agent-sdk';
+
+const session = createSession({
+  model: 'gpt-4o',
+  apiKey: process.env.OPENAI_API_KEY,
+});
+
+// Send message
+await session.send("What is 5 + 3?");
+
+// Stream the response
+for await (const message of session.stream()) {
+  if (message.type === 'assistant') {
+    console.log(message.content);
+  }
+}
+
+// Continue conversation (context preserved)
+await session.send("Multiply that by 2");
+for await (const message of session.stream()) {
+  console.log(message.content);
+}
+
+session.close();
+```
+
+### Advanced Options
+
+```typescript
+const result = await prompt("Analyze the codebase", {
+  model: 'gpt-4o',
+  apiKey: process.env.OPENAI_API_KEY,
+  systemPrompt: "You are a code review assistant.",
+  maxTurns: 15,
+  allowedTools: ['Read', 'Glob', 'Grep'],
+  cwd: './src',
+  env: { NODE_ENV: 'development' },
+  permissionMode: 'default', // 'default' | 'acceptEdits' | 'bypassPermissions' | 'plan'
+});
+```
+
+### With Cancellation
+
+```typescript
+const abortController = new AbortController();
+
+setTimeout(() => abortController.abort(), 30000);
+
+const result = await prompt("Long running analysis...", {
+  model: 'gpt-4o',
+  apiKey: process.env.OPENAI_API_KEY,
+  abortController,
+});
+```
+
+## API Reference
+
+### `prompt(prompt, options)`
+
+Execute a single prompt with the agent using the ReAct loop.
+
+**Parameters:**
+- `prompt` (`string`): User's question or task
+- `options` (`PromptOptions`): Configuration object
+  - `model` (`string`, **required**): Model identifier
+  - `apiKey` (`string`): API key (defaults to env var)
+  - `provider` (`'openai' | 'google'`): Provider (auto-detected if not specified)
+  - `baseURL` (`string`): Base URL for API (OpenAI-compatible)
+  - `maxTurns` (`number`): Maximum conversation turns (default: 10)
+  - `allowedTools` (`string[]`): Allowed tools whitelist
+  - `systemPrompt` (`string`): System prompt
+  - `cwd` (`string`): Working directory (default: `process.cwd()`)
+  - `env` (`Record<string, string>`): Environment variables
+  - `abortController` (`AbortController`): Cancellation support
+  - `permissionMode` (`PermissionMode`): Permission mode
+  - `hooks` (`HooksConfig`): Event hooks configuration
+
+**Returns:** `Promise<PromptResult>`
+- `result` (`string`): Final result text
+- `duration_ms` (`number`): Execution time in milliseconds
+- `usage` (`object`): Token usage statistics
+
+### `createSession(options)` / `resumeSession(id, options)`
+
+Create or resume a persistent conversation session.
+
+**Methods:**
+- `send(message: string): Promise<void>`
+- `stream(): AsyncGenerator<SDKMessage>`
+- `close(): void`
+
+## Built-in Tools
+
+| Tool | Description | Parameters |
+|------|-------------|------------|
+| `Read` | Read file contents, supports images | `file_path`, `offset?`, `limit?` |
+| `Write` | Write content to a file | `file_path`, `content` |
+| `Edit` | Edit file with search/replace | `file_path`, `old_string`, `new_string` |
+| `Bash` | Execute shell commands | `command`, `timeout?`, `run_in_background?` |
+| `Glob` | Find files matching patterns | `pattern`, `path?` |
+| `Grep` | Search code with regex | `pattern`, `path?`, `output_mode?` |
+| `WebSearch` | Search the web | `query`, `numResults?` |
+| `WebFetch` | Fetch webpage content | `url`, `prompt?` |
+| `Task` | Delegate to subagent (includes task management) | `description`, `prompt`, `subagent_type` |
+
+## Provider Support
+
+| Provider | Status | Models Tested |
+|----------|--------|---------------|
+| OpenAI | ✅ Supported | gpt-4o, gpt-4o-mini, gpt-4 |
+| Google Gemini | ✅ Supported | gemini-2.0-flash, gemini-1.5-flash |
+
+## Architecture
+
+```
+┌─────────────────────────────────────────────────────────────┐
+│                     Open Agent SDK                           │
+├─────────────────────────────────────────────────────────────┤
+│  ┌──────────────┐  ┌──────────────┐  ┌──────────────────┐  │
+│  │   prompt()   │  │   Session    │  │  ReActLoop       │  │
+│  │  (One-shot)  │  │ (Persistent) │  │ (Reason + Act)   │  │
+│  └──────┬───────┘  └──────┬───────┘  └────────┬─────────┘  │
+│         └─────────────────┴───────────────────┘            │
+│                           │                                │
+│         ┌─────────────────┼─────────────────┐              │
+│         ▼                 ▼                 ▼              │
+│  ┌────────────┐   ┌──────────────┐   ┌──────────────┐     │
+│  │  Provider  │   │ ToolRegistry │   │  Permission  │     │
+│  │(OpenAI/    │   │(Read/Write/  │   │   Manager    │     │
+│  │ Google)    │   │ Bash/Web...) │   │(4 modes)     │     │
+│  └────────────┘   └──────────────┘   └──────────────┘     │
+│         │                 │                 │              │
+│         └─────────────────┴─────────────────┘              │
+│                           │                                │
+│                    ┌──────▼──────┐                         │
+│                    │HookManager  │                         │
+│                    │(9 events)   │                         │
+│                    └─────────────┘                         │
+└─────────────────────────────────────────────────────────────┘
+```
+
+## Project Status
+
+**Current Version:** v0.1.0-alpha.0
+
+This project is being developed in public. Follow our progress:
+
+- Twitter: [@octane0411](https://twitter.com/octane0411)
+- Discussions: [GitHub Discussions](../../discussions)
+
+### Roadmap
+
+| Version | Features | Status |
+|---------|----------|--------|
+| v0.1.0-alpha | Core ReAct loop, 17 tools, 3 providers, Session, Hooks, Permissions | ✅ Released |
+| v0.1.0-beta | Structured outputs, File checkpointing, Session forking enhancements | 🚧 In Progress |
+| v0.1.0 | Stable release | 📋 Planned |
+| v0.2.0 | Browser automation, Skill system, Query class | 📋 Planned |
+| v1.0.0 | Full Claude Agent SDK compatibility, Python SDK | 📋 Planned |
+
+## Development
+
+```bash
+# Clone the repository
+git clone https://github.com/Octane0411/open-agent-sdk.git
+cd open-agent-sdk
+
+# Install dependencies
+bun install
+
+# Run tests
+bun test
+
+# Run with coverage
+bun test --coverage
+
+# Type checking
+cd packages/core && npx tsc --noEmit
+
+# Run demo
+GEMINI_API_KEY=your-key bun examples/demo.ts
+```
+
+## Why build this?
+
+Claude Agent SDK is excellent but closed-source. We wanted:
+
+1. **Full transparency** — Open code, free to customize
+2. **Provider independence** — No lock-in to a single vendor
+3. **Lightweight core** — Focused, understandable architecture
+4. **Interview-friendly** — Every design decision is explainable
+
+## Contributing
+
+Contributions welcome. Please read [CONTRIBUTING.md](./CONTRIBUTING.md).
+
+## License
+
+[MIT](./LICENSE) © 2026 Octane0411

package/README.zh.md

@@ -0,0 +1,285 @@
+# Open Agent SDK
+
+[![Build in Public](https://img.shields.io/badge/Build%20in%20Public-blue)](https://twitter.com/octane0411)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![TypeScript](https://img.shields.io/badge/TypeScript-5.0+-blue)](https://www.typescriptlang.org/)
+
+Claude Agent SDK 的开源替代品 —— 轻量、可定制、无供应商锁定。
+
+[English Documentation](./README.md)
+
+---
+
+## 这是什么？
+
+Open Agent SDK 是一个用于构建 AI Agent 的 TypeScript 框架。它提供了与 Claude Agent SDK 类似的开发体验，但完全开源透明，没有供应商锁定。
+
+**核心特性：**
+- **ReAct 循环** —— 观察-思考-行动的自主 Agent 循环
+- **内置工具** —— 文件操作（读/写/编辑）、Shell 执行、代码搜索（Glob/Grep）、网页搜索
+- **流式支持** —— 实时响应流和 Token 使用量追踪
+- **多供应商** —— 支持 OpenAI 和 Google Gemini
+- **会话管理** —— 支持内存和文件存储的持久化对话
+- **权限系统** —— 4 种权限模式（default/acceptEdits/bypassPermissions/plan）
+- **Hooks 框架** —— 事件驱动的可扩展性（9 个钩子事件）
+- **子 Agent 系统** —— 将任务委托给专门的 Agent
+- **类型安全** —— 完整的 TypeScript 支持，严格的类型约束
+- **取消支持** —— 通过 AbortController 中断长时间运行的操作
+
+## 安装
+
+```bash
+npm install open-agent-sdk
+```
+
+或指定包管理器：
+
+```bash
+# npm
+npm install open-agent-sdk
+
+# yarn
+yarn add open-agent-sdk
+
+# pnpm
+pnpm add open-agent-sdk
+
+# bun
+bun add open-agent-sdk
+```
+
+**环境要求：**
+- Bun >= 1.0.0（主要运行环境）
+- Node.js >= 20（需安装 peer dependencies）
+- TypeScript >= 5.0
+
+## 快速开始
+
+### 基础用法
+
+```typescript
+import { prompt } from 'open-agent-sdk';
+
+const result = await prompt("当前目录有哪些文件？", {
+  model: 'gpt-4o',
+  apiKey: process.env.OPENAI_API_KEY,
+});
+
+console.log(result.result);
+console.log(`耗时: ${result.duration_ms}ms`);
+console.log(`Token: ${result.usage.input_tokens} 输入 / ${result.usage.output_tokens} 输出`);
+```
+
+### 使用 Gemini
+
+```typescript
+const result = await prompt("解释量子计算", {
+  model: 'gemini-2.0-flash',
+  provider: 'google',
+  apiKey: process.env.GEMINI_API_KEY,
+});
+```
+
+### 基于会话的对话
+
+```typescript
+import { createSession } from 'open-agent-sdk';
+
+const session = createSession({
+  model: 'gpt-4o',
+  apiKey: process.env.OPENAI_API_KEY,
+});
+
+// 发送消息
+await session.send("5 + 3 等于多少？");
+
+// 流式获取响应
+for await (const message of session.stream()) {
+  if (message.type === 'assistant') {
+    console.log(message.content);
+  }
+}
+
+// 继续对话（上下文会被保留）
+await session.send("将结果乘以 2");
+for await (const message of session.stream()) {
+  console.log(message.content);
+}
+
+session.close();
+```
+
+### 高级选项
+
+```typescript
+const result = await prompt("分析代码库", {
+  model: 'gpt-4o',
+  apiKey: process.env.OPENAI_API_KEY,
+  systemPrompt: "你是一个代码审查助手。",
+  maxTurns: 15,
+  allowedTools: ['Read', 'Glob', 'Grep'],
+  cwd: './src',
+  env: { NODE_ENV: 'development' },
+  permissionMode: 'default', // 'default' | 'acceptEdits' | 'bypassPermissions' | 'plan'
+});
+```
+
+### 取消操作
+
+```typescript
+const abortController = new AbortController();
+
+// 30 秒后取消
+setTimeout(() => abortController.abort(), 30000);
+
+const result = await prompt("长时间运行的分析...", {
+  model: 'gpt-4o',
+  apiKey: process.env.OPENAI_API_KEY,
+  abortController,
+});
+```
+
+## API 参考
+
+### `prompt(prompt, options)`
+
+使用 ReAct 循环执行单个提示。
+
+**参数：**
+- `prompt` (`string`): 用户的问题或任务
+- `options` (`PromptOptions`): 配置对象
+  - `model` (`string`, **必需**): 模型标识符
+  - `apiKey` (`string`): API 密钥（默认从环境变量读取）
+  - `provider` (`'openai' | 'google'`): 供应商（未指定时自动检测）
+  - `baseURL` (`string`): API 基础 URL（OpenAI 兼容）
+  - `maxTurns` (`number`): 最大对话轮数（默认：10）
+  - `allowedTools` (`string[]`): 允许使用的工具白名单
+  - `systemPrompt` (`string`): 系统提示词
+  - `cwd` (`string`): 工作目录（默认：`process.cwd()`）
+  - `env` (`Record<string, string>`): 环境变量
+  - `abortController` (`AbortController`): 取消支持
+  - `permissionMode` (`PermissionMode`): 权限模式
+  - `hooks` (`HooksConfig`): 事件钩子配置
+
+**返回：** `Promise<PromptResult>`
+- `result` (`string`): 最终结果文本
+- `duration_ms` (`number`): 执行时间（毫秒）
+- `usage` (`object`): Token 使用统计
+
+### `createSession(options)` / `resumeSession(id, options)`
+
+创建或恢复持久化对话会话。
+
+**方法：**
+- `send(message: string): Promise<void>`
+- `stream(): AsyncGenerator<SDKMessage>`
+- `close(): void`
+
+## 内置工具
+
+| 工具 | 描述 | 参数 |
+|------|------|------|
+| `Read` | 读取文件内容，支持图片 | `file_path`, `offset?`, `limit?` |
+| `Write` | 写入内容到文件 | `file_path`, `content` |
+| `Edit` | 使用查找替换编辑文件 | `file_path`, `old_string`, `new_string` |
+| `Bash` | 执行 Shell 命令 | `command`, `timeout?`, `run_in_background?` |
+| `Glob` | 查找匹配模式的文件 | `pattern`, `path?` |
+| `Grep` | 使用正则搜索代码 | `pattern`, `path?`, `output_mode?` |
+| `WebSearch` | 网页搜索 | `query`, `numResults?` |
+| `WebFetch` | 获取网页内容 | `url`, `prompt?` |
+| `Task` | 委托给子 Agent（包含任务管理） | `description`, `prompt`, `subagent_type` |
+
+## 供应商支持
+
+| 供应商 | 状态 | 已测试模型 |
+|--------|------|-----------|
+| OpenAI | ✅ 已支持 | gpt-4o, gpt-4o-mini, gpt-4 |
+| Google Gemini | ✅ 已支持 | gemini-2.0-flash, gemini-1.5-flash |
+
+## 架构
+
+```
+┌─────────────────────────────────────────────────────────────┐
+│                     Open Agent SDK                           │
+├─────────────────────────────────────────────────────────────┤
+│  ┌──────────────┐  ┌──────────────┐  ┌──────────────────┐  │
+│  │   prompt()   │  │   Session    │  │  ReActLoop       │  │
+│  │  (单次)      │  │ (持久化)     │  │ (推理 + 行动)    │  │
+│  └──────┬───────┘  └──────┬───────┘  └────────┬─────────┘  │
+│         └─────────────────┴───────────────────┘            │
+│                           │                                │
+│         ┌─────────────────┼─────────────────┐              │
+│         ▼                 ▼                 ▼              │
+│  ┌────────────┐   ┌──────────────┐   ┌──────────────┐     │
+│  │  Provider  │   │ ToolRegistry │   │  Permission  │     │
+│  │(OpenAI/    │   │(Read/Write/  │   │   Manager    │     │
+│  │ Google)    │   │ Bash/Web...) │   │(4 种模式)    │     │
+│  └────────────┘   └──────────────┘   └──────────────┘     │
+│         │                 │                 │              │
+│         └─────────────────┴─────────────────┘              │
+│                           │                                │
+│                    ┌──────▼──────┐                         │
+│                    │HookManager  │                         │
+│                    │(9 个事件)   │                         │
+│                    └─────────────┘                         │
+└─────────────────────────────────────────────────────────────┘
+```
+
+## 项目状态
+
+**当前版本：** v0.1.0-alpha.0
+
+本项目正在公开开发中。关注我们的进展：
+
+- Twitter: [@octane0411](https://twitter.com/octane0411)
+- 讨论区: [GitHub Discussions](../../discussions)
+
+### 路线图
+
+| 版本 | 特性 | 状态 |
+|---------|----------|--------|
+| v0.1.0-alpha | 核心 ReAct 循环、17 个工具、3 个供应商、Session、Hooks、权限系统 | ✅ 已发布 |
+| v0.1.0-beta | 结构化输出、文件检查点、会话分叉增强 | 🚧 开发中 |
+| v0.1.0 | 稳定版本 | 📋 计划中 |
+| v0.2.0 | 浏览器自动化、Skill 系统、Query 类 | 📋 计划中 |
+| v1.0.0 | 完整的 Claude Agent SDK 兼容、Python SDK | 📋 计划中 |
+
+## 开发
+
+```bash
+# 克隆仓库
+git clone https://github.com/Octane0411/open-agent-sdk.git
+cd open-agent-sdk
+
+# 安装依赖
+bun install
+
+# 运行测试
+bun test
+
+# 覆盖率测试
+bun test --coverage
+
+# 类型检查
+cd packages/core && npx tsc --noEmit
+
+# 运行演示
+GEMINI_API_KEY=your-key bun examples/demo.ts
+```
+
+## 为什么构建这个项目？
+
+Claude Agent SDK 非常优秀，但是闭源的。我们想要：
+
+1. **完全透明** —— 代码开源，可自由定制
+2. **供应商独立** —— 不被单一供应商锁定
+3. **轻量核心** —— 专注、易理解的架构
+4. **面试友好** —— 每个设计决策都可解释
+
+## 贡献
+
+欢迎贡献。请阅读 [CONTRIBUTING.md](./CONTRIBUTING.md)。
+
+## 许可证
+
+[MIT](./LICENSE) © 2026 Octane0411

package/dist/agent/agent-definition.d.ts

@@ -0,0 +1,107 @@
+/**
+ * AgentDefinition types and validation
+ * Aligned with Claude Agent SDK
+ */
+import { z } from 'zod';
+/**
+ * Permission modes for agent execution
+ */
+export type PermissionMode = 'default' | 'acceptEdits' | 'bypassPermissions' | 'plan';
+/**
+ * Model identifiers for subagent
+ * - 'sonnet' | 'opus' | 'haiku': Specific model presets
+ * - 'inherit': Inherit parent agent's model
+ */
+export type ModelIdentifier = 'sonnet' | 'opus' | 'haiku' | 'inherit';
+/**
+ * Zod schema for AgentDefinition validation
+ */
+export declare const AgentDefinitionSchema: z.ZodObject<{
+    /** Natural language description of when to use this agent */
+    description: z.ZodString;
+    /** List of allowed tools (omit to inherit all from parent) */
+    tools: z.ZodOptional<z.ZodArray<z.ZodString, "many">>;
+    /** System prompt for the agent */
+    prompt: z.ZodString;
+    /** Model to use ('inherit' or omit to use parent's model) */
+    model: z.ZodOptional<z.ZodEnum<["sonnet", "opus", "haiku", "inherit"]>>;
+    /** Maximum turns for the agent (omit to use parent's maxTurns) */
+    maxTurns: z.ZodOptional<z.ZodNumber>;
+    /** Permission mode (omit to use parent's permissionMode) */
+    permissionMode: z.ZodOptional<z.ZodEnum<["default", "acceptEdits", "bypassPermissions", "plan"]>>;
+}, "strip", z.ZodTypeAny, {
+    description: string;
+    prompt: string;
+    permissionMode?: "default" | "acceptEdits" | "bypassPermissions" | "plan" | undefined;
+    model?: "haiku" | "inherit" | "sonnet" | "opus" | undefined;
+    tools?: string[] | undefined;
+    maxTurns?: number | undefined;
+}, {
+    description: string;
+    prompt: string;
+    permissionMode?: "default" | "acceptEdits" | "bypassPermissions" | "plan" | undefined;
+    model?: "haiku" | "inherit" | "sonnet" | "opus" | undefined;
+    tools?: string[] | undefined;
+    maxTurns?: number | undefined;
+}>;
+/**
+ * Agent definition type
+ * Defines the configuration for a subagent
+ */
+export type AgentDefinition = z.infer<typeof AgentDefinitionSchema>;
+/**
+ * Collection of agent definitions
+ * Key is the subagent_type identifier
+ */
+export type AgentDefinitions = Record<string, AgentDefinition>;
+/**
+ * Validate an agent definition
+ * @param definition The definition to validate
+ * @returns The validated definition
+ * @throws ZodError if validation fails
+ */
+export declare function validateAgentDefinition(definition: unknown): AgentDefinition;
+/**
+ * Safely validate an agent definition
+ * @param definition The definition to validate
+ * @returns Object with success flag and data or error
+ */
+export declare function safeValidateAgentDefinition(definition: unknown): {
+    success: true;
+    data: AgentDefinition;
+} | {
+    success: false;
+    error: z.ZodError;
+};
+/**
+ * Create a minimal agent definition with required fields
+ * @param description When to use this agent
+ * @param prompt System prompt for the agent
+ * @returns AgentDefinition with required fields
+ */
+export declare function createAgentDefinition(description: string, prompt: string): AgentDefinition;
+/**
+ * Check if an agent definition has custom tools
+ * @param definition The agent definition
+ * @returns true if tools are explicitly defined
+ */
+export declare function hasCustomTools(definition: AgentDefinition): boolean;
+/**
+ * Check if an agent definition inherits parent's model
+ * @param definition The agent definition
+ * @returns true if model is 'inherit' or undefined
+ */
+export declare function inheritsModel(definition: AgentDefinition): boolean;
+/**
+ * Check if an agent definition has custom maxTurns
+ * @param definition The agent definition
+ * @returns true if maxTurns is explicitly defined
+ */
+export declare function hasCustomMaxTurns(definition: AgentDefinition): boolean;
+/**
+ * Check if an agent definition has custom permission mode
+ * @param definition The agent definition
+ * @returns true if permissionMode is explicitly defined
+ */
+export declare function hasCustomPermissionMode(definition: AgentDefinition): boolean;
+//# sourceMappingURL=agent-definition.d.ts.map