How to Use the Cloudflare MCP in OpenAI Agents SDK

Q: How do I connect the Cloudflare MCP Server to my OpenAI Agents SDK?

Install openai-agents and pass the Vinkius HTTP endpoint to MCPServerStreamableHttpParams. Register this server parameter in the agent constructor's mcpservers list to auto-discover all 25 tools.

Q: Can my OpenAI Agents SDK handle gradual traffic rollouts on Cloudflare?

Yes. Your agent can call createdeployment with a percentage-based strategy to split traffic between Worker versions. This lets the agent monitor performance before committing 100% of traffic.

Q: Does this setup support live log streaming?

Yes. The agent initiates a WebSocket connection using createtailsession to stream console outputs and exceptions. It can then parse these logs to diagnose errors in real time.

Q: What happens to my Cloudflare Worker secrets when queried by the OpenAI Agents SDK?

When the agent runs listsecrets, Cloudflare only returns metadata like names and types. Plaintext values for database passwords or API keys are never exposed back to the SDK or the model, preventing accidental leaks.

Ship code and manage edge resources directly from your OpenAI Agents SDK production deployment.

See Vinkius in Action

Works with every AI agent you already use

…and any MCP-compatible client

MCP Servers - Free for Subscribers

Connect Cloudflare MCP to OpenAI Agents SDK

Create your Vinkius account to connect Cloudflare to OpenAI Agents SDK and route execution through our secure gateway. The platform manages server hosting, runtime updates, and security layers. Configuration requires no manual server provisioning.

GDPR Free for Subscribers

Setup Cloudflare with OpenAI Agents SDK

Ask AI about this MCP

ChatGPT

Claude

Perplexity

Run zero-trust rollouts with your OpenAI Agents SDK

Deploying code updates is handled securely by the `create_deployment` tool without manual terminal commands. The agent uses `list_worker_versions` to inspect recent builds, then triggers the deployment to shift traffic over. If something breaks, the agent handles the rollback by reverting to a stable version ID. Because OpenAI Agents SDK supports strict guardrails, you can restrict these deployment actions to specific, vetted agents. You get full visual tracing of every deployment step inside your OpenAI dashboard, keeping your production environment safe from rogue execution.

Live debugging and secret rotation

Live debugging of production environments starts with the `create_tail_session` tool to capture exceptions instantly. If an API key leaks, the agent swaps it out via `create_secret` and cleans up with `delete_secret` immediately. You don't need to log into a browser dashboard to handle emergencies. Using this MCP Server alongside OpenAI's multi-agent setup means one specialized agent can monitor logs while another updates the environment variables. This isolates sensitive credentials so only the designated security agent ever handles secret values.

Dynamic database querying and cache clearing

Executing SQL queries at the edge is done using the `query_d1` tool to pull configuration tables or run migrations. The agent queries your databases, then purges the CDN cache using `purge_cache` to ensure users see the updates instantly. It cuts out the middleman entirely. You don't have to write custom wrappers for database access. The OpenAI Agents SDK auto-discovers these tools, allowing your agent to read KV keys using `get_kv_key` and verify active routes via `list_worker_routes` in the same execution loop.

Setup guide

Set up Cloudflare MCP in OpenAI Agents SDK

Prerequisites

Python 3.10+ installed
openai-agents package (pip install openai-agents)
Active Vinkius subscription with a valid endpoint token

1

Install the SDK

Run pip install openai-agents to install the OpenAI Agents SDK. The MCP integration is built-in — no extra dependencies needed.
2

Connect via SSE transport

Use MCPServerSse with your Vinkius endpoint URL. Replace [YOUR_TOKEN_HERE] with your token from cloud.vinkius.com. The SDK auto-discovers all Cloudflare tools at runtime.
3

Create your Agent

Pass the MCP to Agent(mcp_servers=[server]). The agent receives Cloudflare tools as native definitions — JSON schemas resolve automatically.
4

Run the agent

Call Runner.run(agent, prompt) to execute. The agent invokes the appropriate Cloudflare tools and returns structured results. Copy the full example on the right to get started.

agent.py

import asyncio
from agents import Agent, Runner
from agents.mcp import MCPServerSse

async def main():
    async with MCPServerSse(
        url="https://edge.vinkius.com/[YOUR_TOKEN_HERE]/mcp"
    ) as server:
        agent = Agent(
            name="Cloudflare Agent",
            instructions="You have access to Cloudflare tools.",
            mcp_servers=[server],
        )
        result = await Runner.run(agent, "List recent transactions")
        print(result.final_output)

asyncio.run(main())

Get your connection token →

Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Cloudflare. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.

Why Choose Vinkius

Vinkius connects your tools to AI with real-time monitoring and automatic cost savings — all from one dashboard.

Connect Cloudflare now

Real-time monitoring

Live

visibility into every interaction

Connect your favorite tools to your AI and see exactly what's happening — every request, every response, in real time.

Built-in savings

60%

lower AI costs

Vinkius compresses data between your apps and your AI automatically. Lower bills every month — no configuration required.

Single dashboard

One

place for every integration

Every tool your AI connects to, managed from a single screen. One account, complete control.

Common questions about Cloudflare MCP in OpenAI Agents SDK

Install `openai-agents` and pass the Vinkius HTTP endpoint to `MCPServerStreamableHttpParams`. Register this server parameter in the agent constructor's `mcp_servers` list to auto-discover all 25 tools.

Yes. Your agent can call `create_deployment` with a percentage-based strategy to split traffic between Worker versions. This lets the agent monitor performance before committing 100% of traffic.

You define system prompts and agent boundaries within your SDK setup. Keep destructive tools like `delete_worker` restricted to a high-privilege agent that requires manual human approval before running.

Yes. The agent initiates a WebSocket connection using `create_tail_session` to stream console outputs and exceptions. It can then parse these logs to diagnose errors in real time.

When the agent runs `list_secrets`, Cloudflare only returns metadata like names and types. Plaintext values for database passwords or API keys are never exposed back to the SDK or the model, preventing accidental leaks.

Use it with your favorite AI tools

Connect this server to Cursor, Claude, VS Code, and more.

OpenAI Agents SDK sdk-python

Google ADK sdk-python

Pydantic AI sdk-python

Vercel AI SDK sdk-typescript