loaditout.ai
SkillsPacksTrendingLeaderboardAPI DocsBlogSubmitRequestsCompareAgentsXPrivacyDisclaimer
{}loaditout.ai
Skills & MCPPacksBlog

aixandria

MCP Tool

gvonness-apolitical/aixandria

AI executive assistant dashboard — visual interface for a Claude-powered assistant with structural human-in-the-loop safety, MCP proxy isolation, and curated knowledge base

Install

$ npx loaditout add gvonness-apolitical/aixandria

Platform-specific configuration:

.claude/settings.json
{
  "mcpServers": {
    "aixandria": {
      "command": "npx",
      "args": [
        "-y",
        "aixandria"
      ]
    }
  }
}

Add the config above to .claude/settings.json under the mcpServers key.

About

Aixandria

A visual dashboard for the Apolitical AI executive assistant. Aixandria makes the assistant tangible — showing its workflows, knowledge base, and safety model in a clean, intuitive interface.

The assistant already exists as a CLI-based Claude Code project with 40+ skills, 8 MCP integrations, and a curated AI Library. Aixandria wraps it in a Next.js dashboard with Docker-based isolation and structural human-in-the-loop safety.

Architecture
┌──────────────────────────────┐
│  Aixandria Dashboard         │  Next.js on host
│  ┌────────────────────────┐  │
│  │ Action Queue (SQLite)  │  │  pending/approved/rejected/failed
│  └────────┬───────────────┘  │
│           │ approved          │
│  ┌────────┴───────────────┐  │
│  │ Action Executor        │  │  deterministic code, traditional APIs
│  └────────────────────────┘  │
└──────┬───────────────────────┘
       │ HTTP (HMAC-signed)
┌──────┴───────────────────────┐
│  Docker Container             │
│  Claude Code (headless)       │  READ-ONLY MCP access
│  Channel Server               │  propose_action / reply / post_status
└──────┬───────────────────────┘
       │ HTTP
┌──────┴───────────────────────┐
│  MCP Proxy (host-side)        │  holds credentials, proxies reads only
│  Spawns MCP servers as        │  write tools stripped from tool surface
│  child processes               │
└───────────────────────────────┘

Key principle: Claude is structurally read-only on external systems. Write operations (send email, post Slack, create ticket) flow through the Action Queue: Claude proposes, the human reviews, deterministic code executes.

Project Structure
aixandria/
  dashboard/          Next.js 15 web app (Week 1+)
  mcp-proxy/          Host-side HTTP server — spawns and proxies MCP servers
  mcp-bridge/         Container-side MCP server — bridges proxy over HTTP
  channel-server/     Container-side MCP server — propose_action, reply, post_status
  docker/             Dockerfiles, entr

Tags

ai-assistantclaudedockerhuman-in-the-loopmcpnextjs

Reviews

Loading reviews...

Quality Signals

0
Installs
Last updated23 days ago
Security: AREADME

Safety

Risk Levelmedium
Data Access
read
Network Accessnone

Details

Sourcegithub-crawl
Last commit3/25/2026
View on GitHub→

Embed Badge

[![Loaditout](https://loaditout.ai/api/badge/gvonness-apolitical/aixandria)](https://loaditout.ai/skills/gvonness-apolitical/aixandria)