# Kenki MCP Agent Setup

Kenki exposes a remote MCP server at:

`https://kenki.ai/mcp`

Users create an agent token from:

`https://kenki.ai/connect-ai.html`

## For AI Agents

When a user gives you a Kenki MCP config, treat it as an explicit request to connect their current agent/client to Kenki.

Do this:

1. Add the provided JSON under the user's MCP server configuration.
2. Keep the bearer token private. Do not repeat it in chat, logs, commits, screenshots, or summaries.
3. Reload or restart MCP tools if your client requires it.
4. Verify the connection by listing MCP tools for the `kenki` server.
5. If tools are available, tell the user Kenki is connected.
6. If you cannot edit MCP config automatically, give the exact file/menu path the user should use and paste only the config shape, with the token redacted unless the user explicitly asks to see it again.

Expected config shape:

```json
{
  "mcpServers": {
    "kenki": {
      "type": "http",
      "url": "https://kenki.ai/mcp",
      "headers": {
        "Authorization": "Bearer <KENKI_AGENT_TOKEN>",
        "MCP-Protocol-Version": "2024-11-05"
      }
    }
  }
}
```

## Suggested User Prompt

Paste this prompt into OpenClaw, Claude Code, Cursor, Cline, Continue, Zed, or another MCP-capable agent along with the config from Kenki:

```text
Connect this Kenki MCP server to your MCP configuration now. Keep the bearer token private. After adding it, reload MCP tools if needed, list the Kenki tools to verify the connection, and tell me whether it worked. If you cannot edit MCP config automatically, tell me the exact place to paste it.
```

## Security

Kenki tokens act on the user's health data according to the scopes selected when the token was created. If a token was pasted into a public or group chat, the user should revoke it in Kenki and create a new one.