OpenAI Integration

Endpoint

OpenAI integrations connect to the same Pixloop MCP endpoint. Responses API remote MCP calls use an OAuth access token in the authorization field. Agents SDK Streamable HTTP clients can pass your Pixloop API key as an Authorization header.

https://mcp.pixloop.ai/mcp

Create or copy an API key in Pixloop Studio under Account Settings → API Keys. You only need this for SDK/server-to-server use or API-key fallback setup.

Responses API

Pass type: "mcp" in the tools array and provide a Pixloop OAuth access token in the authorization field.

from openai import OpenAI

client = OpenAI()

response = client.responses.create(
    model="gpt-5",
    tools=[{
        "type": "mcp",
        "server_label": "pixloop",
        "server_url": "https://mcp.pixloop.ai/mcp",
        "authorization": "YOUR_PIXLOOP_OAUTH_ACCESS_TOKEN"
    }],
    input="Create three static ad variants for my brand and return CDN URLs."
)

print(response.output_text)

Agents SDK — Python

Use MCPServerStreamableHttp from the openai-agents package to give any agent persistent access to Pixloop tools across turns.

pip install openai-agents

from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

pixloop = MCPServerStreamableHttp(
    name="Pixloop",
    params={
        "url": "https://mcp.pixloop.ai/mcp",
        "headers": {"Authorization": "Bearer YOUR_PIXLOOP_API_KEY"},
    },
)

agent = Agent(
    name="Marketing agent",
    model="gpt-5",
    instructions=(
        "You are a creative marketing agent. Use Pixloop tools to "
        "generate brand-aware ads, social posts, and campaign assets."
    ),
    mcp_servers=[pixloop],
)

result = Runner.run_sync(
    agent,
    "Load my brand context, then create a product video concept."
)
print(result.final_output)

Agents SDK — Node.js / TypeScript

npm install @openai/agents

import { Agent, run } from "@openai/agents";
import { MCPServerStreamableHttp } from "@openai/agents";

const pixloop = new MCPServerStreamableHttp({
  name: "Pixloop",
  url: "https://mcp.pixloop.ai/mcp",
  requestInit: {
    headers: { Authorization: `Bearer ${process.env.PIXLOOP_API_KEY}` },
  },
});

const agent = new Agent({
  name: "Marketing agent",
  model: "gpt-5",
  instructions:
    "You are a creative marketing agent. Use Pixloop to generate " +
    "brand-aware ads, manage media, and publish approved posts.",
  mcpServers: [pixloop],
});

const result = await run(
  agent,
  "Create an Instagram carousel for my brand launch."
);
console.log(result.finalOutput);

ChatGPT custom MCP app / connector

ChatGPT discovers Pixloop's OAuth metadata automatically when you add the remote MCP server through ChatGPT settings. Use the server endpoint:

https://mcp.pixloop.ai/mcp

Sign in with your Pixloop account to grant access. Full write-capable MCP tool access may require ChatGPT Developer Mode, depending on your plan and workspace settings.

Available tools

All 11 Pixloop tools are available through the Responses API and Agents SDK MCP integrations. ChatGPT custom MCP app availability depends on your plan, workspace settings, and Developer Mode access. The agent discovers tools through the MCP tools/list method automatically.