# ChatGPT + Render: Manage your Render apps from ChatGPT.

[ChatGPT](https://chatgpt.com) works with Render through the [Render MCP server](/docs/mcp-server), added as a native ChatGPT app — the app type OpenAI now submits and publishes as a ChatGPT *plugin*. With it enabled, ChatGPT can create services, spin up Render Postgres databases, run read-only SQL queries, and pull logs and metrics from a normal chat. Render runs your apps on managed infrastructure (real Postgres with HA and read replicas, persistent disks, background workers, no cold starts, and `render.yaml` Blueprints), so ChatGPT has stable, production-grade surfaces to work with.

## Setup contract

- Auth: API key
- Restart required: No
- Verify: `list_workspaces`

## Agent details

- Agent: ChatGPT
- Vendor: OpenAI
- Docs: [ChatGPT docs](https://help.openai.com/en/articles/12584461-developer-mode-apps-and-full-mcp-connectors-in-chatgpt-beta)

## Quick start

1. *Use a qualifying ChatGPT plan.* Developer mode and custom apps are available on Plus, Pro, Business, and Enterprise plans. They aren't available on the free plan.

2. *Create a Render API key.* Generate one in your [account settings](https://dashboard.render.com/u/settings#api-keys) and keep it handy.

3. *Turn on developer mode.* Open *Settings → Security and login* and enable *Developer mode*. If the toggle is unavailable, a workspace admin needs to allow it. (Menu labels vary by plan and rollout.)

4. *Add the Render app.* Open *Settings → Plugins* (or go to [chatgpt.com/plugins](https://chatgpt.com/plugins)), click the *+* button to create an app, set the MCP server URL to `https://mcp.render.com/mcp`, choose *API key* authentication, and paste your Render API key.

   ```text
   https://mcp.render.com/mcp
   ```

5. *Enable it in a chat.* Start a new conversation, open the *+* (apps) menu in the composer, and turn on *Render*. Apps are off by default in each new chat, and ChatGPT asks you to confirm each action it takes.

6. *Set your Render workspace*, then try a prompt like *"List my Render services and show the latest deploy status for each."*

> Once Render's app is published in the ChatGPT apps directory, you'll be able to add it directly — no developer mode required.

## Quick links

- [Render MCP docs](/docs/mcp-server)
- [MCP Server on GitHub](https://github.com/render-oss/render-mcp-server)
- [ChatGPT developer mode docs](https://help.openai.com/en/articles/12584461-developer-mode-apps-and-full-mcp-connectors-in-chatgpt-beta)
- [Codex + Render](/agents/codex)

## MCP configuration

The Render MCP server is hosted at `https://mcp.render.com/mcp`. ChatGPT connects to it as an app configured in the ChatGPT UI — there's no local config file to edit.

- MCP install command: `Settings → Plugins → +`
- OAuth: Not supported yet

When you add the app, choose the *API key* authentication mode and paste your [Render API key](https://dashboard.render.com/u/settings#api-keys). ChatGPT sends it as an `Authorization: Bearer` header on every request.

> The Render MCP server currently uses *API-key authentication*. Developer mode and custom apps require a paid ChatGPT plan (Plus, Pro, Business, or Enterprise). OAuth support is on the Render roadmap.

For the full tool reference and troubleshooting, see the [MCP server docs](/docs/mcp-server) and the [Render MCP server repo](https://github.com/render-oss/render-mcp-server).

## Example prompts

- *Deploy:* Deploy an example Flask web service on Render from https://github.com/render-examples/flask-hello-world.
- *Data:* Create a Render Postgres database named user-db with 5 GB of storage.
- *Query:* Using my Render database, tell me which items were most frequently bought together.
- *Debug:* Pull the most recent error-level logs for my API service and summarize the likely cause.
- *Metrics:* What was the busiest traffic day for my web service this month?
- *Config:* List the environment variables on my web service and flag anything that looks unset.

## Common workflows

### Spin up a new service

- Ask ChatGPT to create a web service, static site, or cron job from a public Git repo.
- Review the proposed name, plan, and environment variables.
- Confirm the tool call, and let the Render app create it in your active workspace.
- Check the first deploy's status and logs.

### Create and query a database

- Ask ChatGPT to create a Render Postgres database and pick a plan.
- Wire its connection string into a service's environment variables.
- Run read-only SQL queries to explore data and answer questions.

### Investigate logs and metrics

- Ask ChatGPT for the most recent error-level logs for a service.
- Have it summarize CPU, memory, and response metrics over a time range.
- Decide on next steps, then apply supported changes like environment variable updates.

## Tips for success

> - *Re-enable the Render app in each new chat.* Apps are off by default in new conversations — turn Render on again from the composer's *+* menu.
> - *Confirm tool calls when prompted.* ChatGPT pauses and waits for your approval before it runs a Render action.
> - *Set your workspace first* so ChatGPT acts against the right account resources.
> - *Developer mode has full read/write MCP access.* The Render MCP server can create supported services and update environment variables, but it can't trigger deploys, change scaling, or delete resources — use the [Render Dashboard](https://dashboard.render.com), [API](/docs/api), or [CLI](/docs/cli) for those.
> - *For repo-aware deploys and code edits, use [Codex with Render](/agents/codex).*

## FAQ

###### Do I need a paid ChatGPT plan?

Yes. Developer mode and custom apps are available on Plus, Pro, Business, and Enterprise plans. The free plan doesn't include the developer-mode toggle or custom apps.

###### How does ChatGPT authenticate to Render?

When you add the Render app, choose the *API key* authentication mode and paste your [Render API key](https://dashboard.render.com/u/settings#api-keys). ChatGPT sends it as an `Authorization: Bearer` header on each request to the MCP server.

###### Why did the Render app stop working in a new chat?

Apps aren't automatically active in new conversations. Re-enable Render from the composer's *+* menu at the start of each chat.

###### Can ChatGPT deploy from my local repo or edit code?

Not directly — ChatGPT works through the hosted MCP server, not your working directory. For repo-aware deploys, Blueprint edits, and code fixes, use [Codex with Render](/agents/codex).

## Troubleshooting

### There's no button to add an app

The *+* to create an app in *Settings → Plugins* only appears after you enable developer mode. Turn it on under *Settings → Security and login* first. If the toggle is missing, ask your workspace admin to allow developer mode for your account.

### The app is saved but nothing happens in chat

A saved app isn't active until you enable it for the conversation. Start a new chat, then turn Render on from the composer's *+* menu. Also confirm any tool call ChatGPT proposes — it waits for your approval before acting.

### Authentication errors from the Render MCP

Make sure you selected the *API key* authentication mode and pasted a valid Render key with access to the target workspace. Tokens embedded in the URL are rejected — the key belongs in the auth field, not the server URL.

## Community

*Render MCP Server*

Open-source MCP server source, install instructions, and tool reference.

*Render Skills*

The full skill catalog with `SKILL.md` files for every supported task.

*Render Discord*

Get help from the Render team and community in real time.

*Render on GitHub*

Open-source repos, examples, and the platform changelog.

- [Start a new project](https://dashboard.render.com)
- [Read the docs](/docs)