Vercel Adapter

API reference for the Vercel handle_vercel adapter

The Vercel adapter enables deploying agents as serverless functions on Vercel’s edge platform.

Import

from ai_query.adapters.vercel import handle_vercel

---

handle_vercel

The handle_vercel function bridges Vercel’s HTTP handler to the agent system.

Function Signature

def handle_vercel(agent: Agent, handler: BaseHTTPRequestHandler) -> None

Parameters

agent · Agent · required

The agent instance to handle requests.

handler · BaseHTTPRequestHandler · required

Vercel’s HTTP handler instance.

Usage Examples

Basic Vercel API Route

# api/index.py
from http.server import BaseHTTPRequestHandler
from ai_query.agents import Agent
from ai_query.adapters.vercel import handle_vercel
from ai_query.providers import openai

# Create agent
agent = Agent(
    "vercel-bot",
    model=openai("gpt-4o"),
    system="You are a Vercel-hosted assistant."
)

class handler(BaseHTTPRequestHandler):
    def do_POST(self):
        handle_vercel(agent, self)

    def do_GET(self):
        handle_vercel(agent, self)

Multiple Endpoints

# api/chat.py
from http.server import BaseHTTPRequestHandler
from ai_query.agents import Agent
from ai_query.adapters.vercel import handle_vercel
from ai_query.providers import openai

chat_agent = Agent("chat-agent", model=openai("gpt-4o"))

class handler(BaseHTTPRequestHandler):
    def do_POST(self):
        handle_vercel(chat_agent, self)

# api/research.py
research_agent = Agent("research-agent", model=openai("gpt-4o"))

class handler(BaseHTTPRequestHandler):
    def do_POST(self):
        handle_vercel(research_agent, self)

With Environment Variables

import os
from http.server import BaseHTTPRequestHandler
from ai_query.agents import Agent
from ai_query.adapters.vercel import handle_vercel
from ai_query.providers import openai

agent = Agent(
    "vercel-agent",
    model=openai(os.getenv("OPENAI_MODEL", "gpt-4o")),
    system=os.getenv("SYSTEM_PROMPT", "You are helpful.")
)

class handler(BaseHTTPRequestHandler):
    def do_POST(self):
        handle_vercel(agent, self)

---

Project Structure

my-vercel-app/
├── api/
│   ├── index.py          # Main agent
│   ├── chat.py           # Chat agent
│   └── research.py       # Research agent
├── agents/
│   ├── __init__.py
│   ├── chat_agent.py
│   └── research_agent.py
├── package.json
└── vercel.json

# api/index.py
from http.server import BaseHTTPRequestHandler
from agents.chat_agent import create_chat_agent
from ai_query.adapters.vercel import handle_vercel

agent = create_chat_agent()

class handler(BaseHTTPRequestHandler):
    def do_POST(self):
        handle_vercel(agent, self)
    def do_GET(self):
        handle_vercel(agent, self)

---

Error Handling

The Vercel adapter returns a plain text error response:

"Invalid JSON body"

For custom error handling, wrap the agent in your handler class.