OpenAI
The official Synthflow plugin lets you manage your workspace with natural-language prompts in ChatGPT and Codex, through the Synthflow MCP server.
You can also bring your own API key so that Synthflow runs supported models with your own OpenAI credentials. This is independent from the plugin: it changes which credentials power model access for your agents, not how you manage your workspace.
What you can do
Ask ChatGPT or Codex to:
- Build and update agents: Create agents, draft and refine prompts, publish drafts, and compare versions before you go live.
- Manage calls: Place test calls, review recent calls, and surface summaries or negative-sentiment moments when you ask.
- Manage knowledge: Create knowledge bases, add items from files or URLs, and attach them to the agents that should use them.
- Run actions and simulations: Scaffold and test actions, then run simulations to validate behavior before launch.
The MCP Server page lists every supported tool.
Plugin installation
You sign in through Synthflow, so there is no API key to copy or manage by hand. Setup differs slightly between ChatGPT and Codex.
ChatGPT
Open the Synthflow plugin in ChatGPT and install it.
When ChatGPT shows Finish setting up Synthflow, connect the connector that matches your workspace region. The list includes Synthflow EU, Synthflow Global, and Synthflow US. Connect only the one that matches.
You are redirected to Synthflow to log in, where you select the workspace you want to manage.
After you select the workspace, select Allow access on the next screen. You can then start prompting.
Codex
Add the server with the URL for your region, then sign in:
The browser opens the Synthflow sign-in. Select the workspace and select Allow access, the same way as in ChatGPT. The Codex CLI and IDE extension share this configuration.
For CI and scheduled runs, use API key authentication instead of the browser sign-in.
Example prompts
The MCP Server page has more example prompts for reporting, testing, and debugging.
Security considerations
Call transcripts, knowledge base documents, and fetched web pages can contain text that tries to steer ChatGPT or Codex into an unintended action, known as prompt injection. Keep write and delete tools behind approval and review each call. Changes and guardrails covers the safeguards on the Synthflow side.
Bring your own key
Connect your OpenAI API key when you want Synthflow to run supported models with your own credentials instead of a shared key. This is separate from the Synthflow plugin above: it changes which credentials power model access for your agents, not how you manage your workspace.
Before you connect
Make sure you have:
- An active OpenAI platform account.
- An API key from the OpenAI API keys dashboard.
Generate the key in your OpenAI account settings, then copy it before you return to Synthflow. OpenAI shows the full key value only once.
Connect your API key in Synthflow
After connection, Synthflow routes supported LLM traffic through your OpenAI account using your API key. Billing and usage limits follow the terms of your OpenAI account.
FAQ
What is the difference between the Synthflow plugin and BYOK?
The Synthflow plugin lets you manage your workspace with natural-language prompts in ChatGPT and Codex. BYOK is separate: it tells Synthflow to run supported models with your own OpenAI credentials. You can use any combination.
Which connector should I connect in ChatGPT?
Connect Synthflow Global, Synthflow US, or Synthflow EU to match your workspace region. Check the region under Settings → Workspace Settings → Preferences → Customer Region. Connecting the wrong one stops the plugin from working.
How is this different from Aurora?
Where do I generate an OpenAI API key?
Create one in the OpenAI API keys dashboard. Copy the key before you leave the page.
Why do I not see OpenAI in my workspace integrations?
The OpenAI API key integration appears only when BYOK access is enabled for your account. Contact your account team if you need this option added.
How is model usage billed?
ChatGPT and Codex run in your own OpenAI account, so model and subscription charges are billed by OpenAI according to your plan. Synthflow bills only for the workspace usage your actions generate, such as calls, under your existing Synthflow plan.
Can I use Azure instead of OpenAI?
Yes. If you run models through Azure OpenAI, connect Azure and provide your Azure endpoint, API version, and API key.