TypeScript Executor

Execute TypeScript code with API interception

Overview

The TypeScriptExecutorProxy intercepts the fetch API and routes calls to isolated environments. Code runs in-process using Node.js.

Basic Usage

import { TypeScriptExecutorProxy } from 'agent-diff';

// Create executor for an environment
const executor = new TypeScriptExecutorProxy(
  'env-123',
  'http://localhost:8000'  // Optional, defaults to localhost
);

// Execute code
const result = await executor.execute(`
  const response = await fetch('https://slack.com/api/chat.postMessage', {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      channel: 'C01GENERAL99',
      text: 'Hello from TypeScript!'
    })
  });
  const data = await response.json();
  console.log(data);
`);

console.log(result.stdout);  // API response
console.log(result.status);  // "success" or "error"

Creating Framework Tools

Vercel AI SDK

import { TypeScriptExecutorProxy, createVercelAITool } from 'agent-diff';
import { generateText } from 'ai';
import { openai } from '@ai-sdk/openai';

const executor = new TypeScriptExecutorProxy(env.environmentId, client.getBaseUrl());
const tool = await createVercelAITool(executor);

const result = await generateText({
  model: openai('gpt-4o'),
  tools: { execute_typescript: tool },
  prompt: "Post 'Hello' to Slack channel #general",
  maxSteps: 5
});

Supported APIs

The TypeScript executor intercepts:

• fetch: Global fetch API
• node-fetch: (If imported)

Execution Result

const result = await executor.execute(code);

// Result structure:
interface ExecutionResult {
  status: 'success' | 'error';
  stdout: string;     // Captured console.log output
  stderr: string;     // Error output if any
  exitCode?: number;  // Process exit code
  error?: string;     // Error message if any
}

Error Handling

const result = await executor.execute(code);

if (result.status === 'error') {
  console.error(`Execution failed: ${result.stderr}`);
} else {
  console.log(`Output: ${result.stdout}`);
}

Example: Slack API Calls

const result = await executor.execute(`
  // List channels
  const channelsRes = await fetch('https://slack.com/api/conversations.list');
  const channels = await channelsRes.json();
  console.log('Channels:', channels.channels.map(c => c.name));

  // Post message
  const msgRes = await fetch('https://slack.com/api/chat.postMessage', {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      channel: 'C01GENERAL99',
      text: 'Hello!'
    })
  });
  const msg = await msgRes.json();
  console.log('Message sent:', msg.ok);
`);

Example: Linear GraphQL

const result = await executor.execute(`
  const query = \`
    mutation {
      issueCreate(input: {
        title: "Fix bug"
        teamId: "team-123"
      }) {
        success
        issue { id title }
      }
    }
  \`;

  const response = await fetch('https://api.linear.app/graphql', {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({ query })
  });

  console.log(await response.json());
`);

Configuration Options

const executor = new TypeScriptExecutorProxy(
  environmentId: string,
  baseUrl?: string,      // Default: http://localhost:8000
  token?: string         // Optional auth token
);

Async/Await Support

The TypeScript executor fully supports async/await:

const result = await executor.execute(`
  async function main() {
    const res1 = await fetch('https://slack.com/api/conversations.list');
    const channels = await res1.json();
    
    for (const channel of channels.channels.slice(0, 3)) {
      await fetch('https://slack.com/api/chat.postMessage', {
        method: 'POST',
        headers: { 'Content-Type': 'application/json' },
        body: JSON.stringify({
          channel: channel.id,
          text: 'Hello to ' + channel.name
        })
      });
    }
  }
  
  await main();
`);

Next Steps

Python Executor

Execute Python code

Bash Executor

Execute shell commands