feat(llm-sdk): add runtime.generate() for non-streaming responses

[Sahil5963]

Summary

• Add runtime.generate() method for clean non-streaming API (similar to Vercel AI SDK's generateText())
• Add GenerateResult class with toResponse() for CopilotChat-compatible format
• Update Express demo, NextJS demo, and create-ai-copilot templates

New API

const result = await runtime.generate(body);

// Raw access
result.text         // Generated text
result.messages     // All messages
result.toolCalls    // Tool calls
result.success      // Boolean

// CopilotChat format
result.toResponse() // { success, content, messages, ... }

API Comparison

Mode	Raw Format	CopilotChat Format
Streaming	stream().collect()	stream().toResponse()
Non-streaming	generate().text	generate().toResponse()

Usage Examples

Next.js:

export async function POST(req: Request) {
  const result = await runtime.generate(await req.json());
  return Response.json(result.toResponse());
}

Express:

app.post('/api/chat/generate', async (req, res) => {
  const result = await runtime.generate(req.body);
  res.json(result.toResponse());
});

Test plan

• Build llm-sdk package
• Test Express demo /api/chat/generate endpoint
• Verify CopilotChat-compatible response format

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Review	Updated (UTC)
copilot-sdk-docs	Ready	Preview, Comment	Jan 27, 2026 2:35pm