Connect Creed to ChatGPT
To connect Creed to ChatGPT, add it as a custom remote MCP connector in ChatGPT's settings, then authorize it once with OAuth. Open Settings, go to Connectors, add a custom or remote MCP server using the URL https://creed.md/mcp, and authorize by clicking Allow on the Creed consent screen while signed in to creed.md. There is no key to paste; the whole handshake is OAuth.
Once connected, ChatGPT can read your Creed profile before it answers, so it starts already knowing your identity, goals, work, and preferences instead of asking you to re-explain them. As it learns something durable about you, it can propose a narrow update you approve.
Creed is one personal context file that every AI reads before it answers. Connector availability in ChatGPT depends on your plan and the current rollout, so the exact menu wording may differ. The setup shape is the same: add the remote MCP server, authorize, and verify by calling read_creed once.
Before you start
Custom and remote MCP connectors in ChatGPT are gated by plan and rollout, so where the option lives can vary. In general terms, look under Settings for a Connectors area that lets you add a custom or remote MCP server. If you do not see it, your plan or the current rollout may not expose it yet.
Step by step
- 1.Open ChatGPT Settings.
- 2.Go to Connectors.
- 3.Choose to add a custom or remote MCP server.
- 4.Enter the Creed server URL: https://creed.md/mcp
- 5.Save, then start the authorization when prompted.
- 6.A browser step opens the Creed consent screen; sign in to creed.md if needed.
- 7.Click Allow to grant ChatGPT access to your Creed.
The URL is the one detail that matters most. Use the Creed MCP endpoint exactly:
https://creed.md/mcpAuthorizing with OAuth
Creed MCP uses OAuth, so there is nothing to copy. When you authorize, you are taken to the Creed consent screen. Make sure you are signed in to creed.md, then click Allow. That grant is what lets ChatGPT read your profile; you do not create or paste any token.
Verify the connection
Confirm ChatGPT can reach Creed before you rely on it.
- Check that the Creed connector appears as connected in your Connectors list.
- In a chat, confirm the Creed tools are available and have ChatGPT call read_creed once.
- Check that your profile sections come back, which confirms the read path works.
How ChatGPT uses your profile
With Creed connected, ChatGPT can read your profile before it answers rather than starting cold. It uses your Identity, Goals, Work, Preferences, and Routines to shape replies, so you stop re-explaining who you are. When it learns something durable, it can propose a narrow update to the relevant section, which you approve.
Updates flow through the creed_* tools and are scoped to specific sections, not wholesale rewrites. Whether an edit applies directly or arrives as a proposal depends on your write policy, which an agent can check with get_write_policy before changing anything.
One profile across tools
The point of connecting Creed to ChatGPT is consistency. The profile ChatGPT reads is the same one your other connected agents read, so your context does not fork per tool. It is plain Markdown you own, with export and delete anytime and no lock-in.
Frequently asked questions
Open ChatGPT Settings, go to Connectors, add a custom or remote MCP server using the URL https://creed.md/mcp, then authorize by clicking Allow on the Creed consent screen while signed in to creed.md.
Custom and remote MCP connector availability depends on your plan and the current rollout. If you do not see the option to add a custom MCP server under Connectors, your plan or the rollout may not expose it yet.
No. Creed MCP uses OAuth. You authorize through the Creed consent screen by clicking Allow, and there is nothing to paste.
Check that the Creed connector shows as connected, then have ChatGPT call read_creed once. If your profile sections come back, the connection works.