@agi-cli/api

@agi-cli/api

Type-safe API client for AGI CLI server, generated from OpenAPI specification using @hey-api/openapi-ts.

Features

• ✅ Type-safe SDK - Fully typed API functions generated from OpenAPI spec
• 🚀 Axios-powered - Uses Axios for reliable HTTP requests with interceptors support
• 📦 Tree-shakeable - Import only what you need
• 🔄 SSE Streaming - Built-in support for Server-Sent Events
• ✨ Runtime validation - Optional schema validation with generated schemas
• 🎯 Auto-generated - Always in sync with the server API

Installation

npm install @agi-cli/api axios
# or
bun add @agi-cli/api axios
# or
pnpm add @agi-cli/api axios

Quick Start

import { client, ask, listSessions } from '@agi-cli/api';

// Configure the client once at app startup
client.setConfig({
  baseURL: 'http://localhost:3000',
});

// Make type-safe API calls
const response = await ask({
  body: {
    prompt: 'Hello, AI!',
    sessionId: 'optional-session-id',
  },
});

if (response.error) {
  console.error('Error:', response.error);
} else {
  console.log('Response:', response.data);
}

// List all sessions
const sessions = await listSessions();
console.log('Sessions:', sessions.data);

Configuration

Basic Configuration

import { client } from '@agi-cli/api';

client.setConfig({
  baseURL: 'http://localhost:3000',
  // Optional: configure timeout
  timeout: 30000,
});

Advanced Configuration with Interceptors

import { client } from '@agi-cli/api';

// Access the underlying Axios instance
client.instance.interceptors.request.use((config) => {
  // Add authentication token
  config.headers.set('Authorization', `Bearer ${getToken()}`);
  return config;
});

client.instance.interceptors.response.use(
  (response) => response,
  (error) => {
    console.error('API Error:', error);
    return Promise.reject(error);
  }
);

Authentication

import { client } from '@agi-cli/api';

// Configure auth token (will be added to requests that require auth)
client.setConfig({
  baseURL: 'http://localhost:3000',
  auth: () => `Bearer ${getToken()}`,
});

API Reference

All SDK functions are auto-generated and fully typed. Import them directly:

import {
  // Session management
  listSessions,
  createSession,
  subscribeSessionStream,
  
  // Messages
  listMessages,
  createMessage,
  
  // Ask endpoint
  ask,
} from '@agi-cli/api';

SSE Streaming

For endpoints that support Server-Sent Events:

import { createSSEStream } from '@agi-cli/api';

const stream = createSSEStream({
  url: 'http://localhost:3000/v1/sessions/session-123/stream',
  onMessage: (event) => {
    console.log('Event:', event);
  },
  onError: (error) => {
    console.error('Stream error:', error);
  },
});

// Close the stream when done
stream.close();

Error Handling

import { ask, isApiError, handleApiError } from '@agi-cli/api';

const response = await ask({
  body: { prompt: 'Hello' },
});

if (response.error) {
  if (isApiError(response.error)) {
    // Handle API errors
    const { status, message } = handleApiError(response.error);
    console.error(`API Error [${status}]:`, message);
  } else {
    // Handle network or other errors
    console.error('Unexpected error:', response.error);
  }
}

Development

Generating the Client

The client is auto-generated from the server's OpenAPI specification:

# Generate from the latest server spec
bun run generate

# Build the package
bun run build

Configuration

The code generation is configured in openapi-ts.config.ts:

import { defineConfig } from '@hey-api/openapi-ts';

export default defineConfig({
  input: './openapi.json',
  output: {
    path: './src/generated',
  },
  plugins: [
    '@hey-api/typescript',     // Generate TypeScript types
    '@hey-api/schemas',        // Generate runtime schemas
    '@hey-api/sdk',            // Generate SDK functions
    '@hey-api/client-axios',   // Use Axios client
  ],
});

Architecture

@agi-cli/api/
├── src/
│   ├── generated/          # Auto-generated files (don't edit!)
│   │   ├── client.gen.ts   # Axios client instance
│   │   ├── sdk.gen.ts      # SDK functions
│   │   ├── types.gen.ts    # TypeScript types
│   │   └── schemas.gen.ts  # Runtime schemas
│   ├── runtime-config.ts   # Client runtime configuration
│   ├── streaming.ts        # SSE utilities
│   ├── utils.ts           # Helper functions
│   └── index.ts           # Public API exports
├── openapi-ts.config.ts   # Code generation config
├── generate.ts            # Generation script
└── build.ts              # Build script

Migration from Legacy Client

If you're migrating from the old Fetch-based client:

Before (Legacy)

import { createApiClient } from '@agi-cli/api';

const client = createApiClient({
  baseUrl: 'http://localhost:3000',
});

const response = await client.ask({
  prompt: 'Hello',
});

After (New Axios Client)

import { client, ask } from '@agi-cli/api';

// Configure once at startup
client.setConfig({
  baseURL: 'http://localhost:3000',
});

// Use SDK functions
const response = await ask({
  body: { prompt: 'Hello' },
});

Resources

• Hey API Documentation
• Axios Client Guide
• Server Package - The API server that this client connects to

License

MIT