This MCP server provides EU AI Act Art-14 runtime oversight for agent actions. It supports controlling outcomes by allowing, flagging, or gating actions to a human, while also generating a receipt for the oversight event.

🛠️ Key Features

EU AI Act Art-14 runtime oversight
Human intervention control modes: allow / flag / gate-to-human
Receipt included with the action oversight

🚀 Use Cases

Runtime monitoring and control of an agent action under EU AI Act Art-14

⚠️ Limitations

Server details provided only cover runtime oversight for agent actions (allow/flag/gate-to-human with receipt); no additional tooling or capabilities are specified.

{ "type": "object", "properties": { "action": { "type": "string" }, "context": { "default": "", "type": "string" } }, "required": [ "action" ], "additionalProperties": false }

Install

Tools (3)

check_actioncheck_action_deterministiclist_charter

Remote endpoint

Streamable HTTP

Hosted server - connect over the network, no local install.

url

https://vf.humanified.dk/mcp

Add to VS Code Add to Cursor

claude_desktop_config.json

json

{
  "mcpServers": {
    "values-firewall": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://vf.humanified.dk/mcp"
      ]
    }
  }
}

Desktop config is stdio-only; this bridges via mcp-remote. Native remote: Settings > Connectors.

action	string	required
context	string	optional

action	string	required
context	string	optional