MCP server

Connect Claude and other MCP clients to Eevo, and the tools they get.

Eevo's MCP server is at https://eevo.io/api/arvo/mcp (streamable HTTP). Add it to Claude, ChatGPT or another MCP client and sign in, or send an API key. Its decisions bill like the API and appear in the decision log. Step by step for each client: Connect your agent to Eevo.

Claude and ChatGPT: sign in

Add a custom connector with the URL above. In Claude: Customize › Connectors › Add custom connector (on Team and Enterprise plans an Owner adds it under Organization settings › Connectors). In ChatGPT: Plugins › Add › Add custom MCP server, with OAuth authentication, then Create as a plugin.
Click Connect (in ChatGPT, Continue to Eevo). Sign in to Eevo, or create an account if you have none; a new account gets an organization with 5 Default mode decisions, each revisable up to 10 times, and 50 Direct mode decisions.
Allow the connection. The client stays signed in: it renews the connection itself, and asks you to sign in again only after 90 days without use.

Decisions made this way are billed to your organization and kept for its connector retention (Data controls in the console), like the API playground's, and the decision log shows who made them and in which app. Whether your colleagues see them follows your sharing setting (Data controls): an organization's default, which each member can change for themselves. To end a connection, remove the connector in the client.

With an API key

For Claude Code, scripts and servers, send a key from API keys as a bearer token. Its retention applies to its decisions, so use one that suits what the agent sends.

Claude Code:

claude mcp add --transport http eevo https://eevo.io/api/arvo/mcp --header "Authorization: Bearer $EEVO_API_KEY"

Claude Desktop, in claude_desktop_config.json:

{
  "mcpServers": {
    "eevo": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://eevo.io/api/arvo/mcp",
        "--header",
        "Authorization: Bearer ${EEVO_API_KEY}"
      ],
      "env": {
        "EEVO_API_KEY": "eevo_sk_…"
      }
    }
  }
}

Other MCP clients:

{
  "mcpServers": {
    "eevo": {
      "type": "http",
      "url": "https://eevo.io/api/arvo/mcp",
      "headers": {
        "Authorization": "Bearer eevo_sk_…"
      }
    }
  }
}

Tools

Tool	What it does
`eevo_decide`	`POST /v1/decisions` as a tool, in either mode.
`eevo_show_decision`	Show the user a decision as an interactive card, once, for the final decision after any revisions. Free.
`eevo_systemone`	Typed questions about a state.
`eevo_report_chosen_decision`	Report what was done. Free.
`eevo_assess`	Record whether the agent agrees, per stakeholder. Free.
`eevo_share_decision`	Share a decision with the organization, or stop sharing it, when the user asks. Free.
`eevo_list_receipts`	Your organization’s payments with their Stripe receipts, like `GET /v1/receipts`. With an API key only. Free.

The tools take the same fields as the API, including metadata, revises, revision_note and share_with_org. In clients that show MCP Apps (Claude, ChatGPT and others), eevo_show_decision's card shows one stakeholder at a time with a picker, and links to the decision in the console. eevo_decide shows no card, so an agent can revise without showing every revision. The card needs the decision's content, so a key with retention "Not stored" gets none. The tool descriptions tell the agent when to call Eevo and how to present the answer, so it needs no other setup.

For agents that read docs

Every page of these docs is also served as Markdown, listed in llms.txt. Point an agent there to let it read the docs itself.