On this page
Remote MCPInstall guide

Connect your AI client to Ethics of AI

Add the Ethics of AI MCP connector in Claude Desktop, on claude.ai, or on mobile, so Claude can drive assessments and read your account on your behalf.

Add the Ethics of AI connector once, in any Claude surface, and it appears everywhere you sign in to the same Anthropic account. This guide covers Claude Desktop, claude.ai, and mobile.

Before you start. You need an Ethics of AI account with at least one purchased certification. Sign up at ethicsofai.io and buy from the pricing page first; the connector only acts on certifications tied to your account. Buying from inside Claude is on the roadmap.

Connector details

The same values work on every surface:

Name: Ethics of AI
URL:  https://mcp.ethicsofai.io/mcp

You sign in with your Ethics of AI email and password.

How one install reaches every surface

Connectors bind to your Anthropic account, not a single device. Add it once and every signed-in Claude surface picks it up:

That is why the mobile section below is "verify it synced" rather than "set it up again".

Claude Desktop

The most fully-supported surface today. Five clicks.

1. Open the Customise panel from the Claude Desktop sidebar.

Customise panel highlighted in the Claude Desktop sidebar

2. Open Connectors and click + next to Personal plugins.

Connectors panel with the Personal plugins add button

3. Click Add custom connector, name it Ethics of AI, and paste the URL above.

4. Authenticate with your Ethics of AI credentials when Claude prompts you.

5. Open a new chat and ask about your account.

Claude on the web (claude.ai)

Settings, then Connectors, then Add custom connector. Same name, same URL, same credentials.

Claude on mobile

Mobile inherits the connector from Desktop or claude.ai. To confirm on iOS:

1. Open Settings → Connectors in the Claude app.

Claude iOS Settings screen showing the Connectors menu item

2. Ethics of AI is already in the list, signed in with the credentials you used elsewhere.

3. Use it. The same MCP tools work in the same conversation flow:

Claude on iOS responding to a certificate download-links query with PDF, PNG, and badge variants pulled from the Ethics of AI MCP

What happens when Claude calls a tool

Plain language in, account data out. The MCP server signs the request as you, hits the Users API, and returns the answer:

The connector only sees what your account already sees, with the same authorisation rules as the website.

Verify it works

Ask Claude something about your account in plain language:

  • What assessments do I have on Ethics of AI?
  • Show me my certifications.
  • What is the status of my Awareness tier?

Claude picks the right MCP tool and answers inside the conversation:

Claude calling list_my_certifications to answer a question about the user's assessments

If Claude suggests opening the website instead, the connector did not authenticate. Re-run step 4 in the relevant surface above.