Ceveto MCP Server

MCP (Model Context Protocol) server for AI agent access to the Ceveto business management API.

Quick Start

uvx --from git+https://github.com/Ceveto/mcp-server ceveto-mcp

Related MCP server: agentforge

Claude Code Setup

Add to your .mcp.json:

{
  "mcpServers": {
    "ceveto": {
      "command": "uvx",
      "args": ["--from", "git+https://github.com/Ceveto/mcp-server", "ceveto-mcp"],
      "env": {
        "CEVETO_MCP_BASE_URL": "https://api.ceveto.com",
        "CEVETO_MCP_USERNAME": "<your-api-username>",
        "CEVETO_MCP_PRIVATE_KEY": "<your-api-private-key>"
      }
    }
  }
}

Or install the Ceveto Claude Plugin for automated setup.

Getting Credentials

1. Log in to your Ceveto dashboard

2. Go to Settings → API Keys

3. Click Create API Key

4. Save the username and private key (shown once)

Environment Variables

How It Works

1. On startup, fetches the OpenAPI schema from the Ceveto API

2. Checks your permissions via /company-api/me/

3. Generates MCP tools for each permitted API endpoint

4. Tools include parameter schemas, descriptions, and permission limits

Docker (SSE mode)

docker build -t ceveto-mcp .
docker run -p 8500:8500 \
  -e CEVETO_MCP_USERNAME=... \
  -e CEVETO_MCP_PRIVATE_KEY=... \
  -e CEVETO_MCP_BASE_URL=https://api.ceveto.com \
  ceveto-mcp

Hosted Instances

Hosted mode (multi-tenant, no credentials needed on server):

ceveto-mcp --hosted --transport sse --port 8500

Security

• Ed25519 signatures — every API request is cryptographically signed

• OAuth 2.1 — browser-based authorization with PKCE

• Permission-filtered tools — only endpoints you have access to become tools

• Method-level filtering — read-only keys don't see write operations

• No secrets stored — credentials only in environment variables