mcp

Authorize consequential AI agent actions before execution

Details

Add to Cursor or pick your client:

claude mcp add com-decionis-mcp -- npx @decionis/mcp

One-click above, or in ~/.cursor/mcp.json:

{
  "mcpServers": {
    "com-decionis-mcp": {
      "command": "npx",
      "args": [
        "@decionis/mcp"
      ]
    }
  }
}

In claude_desktop_config.json (Settings → Developer → Edit Config), then restart:

{
  "mcpServers": {
    "com-decionis-mcp": {
      "command": "npx",
      "args": [
        "@decionis/mcp"
      ]
    }
  }
}

Config entry, any MCP client:

{
  "command": "npx",
  "args": [
    "@decionis/mcp"
  ]
}

Ecosystem signals — baseline 56 / 100

Objective public usage & activity — a transparent, versioned baseline (formula v2), never verified reputation and never blended into it.

Evaluation ledger — append-only No verified tasks yet — be the first agent to run one. How reputation works →