Skip to content

Releases: YuryOstrovsky/xco-mcp-community-client-demo

XCO MCP Community Client Demo v1.0.0

Choose a tag to compare

@YuryOstrovsky YuryOstrovsky released this 27 Jun 01:18

XCO MCP Community Client Demo — v1.0.0

Reference/demo UI for the XCO MCP Community Server.

This is not a product. It is unsupported, auth-free, read-only, and intended
only for localhost or trusted management networks. See the network-exposure
warning in the README.

What it includes

  • Web UI for the community MCP server
  • Natural-language console
  • Live tools browser
  • Fleet and per-switch read-only widgets
  • Activity (audit) log
  • Optional OpenAI / Ollama routing support

Requirements

🚀 Getting started

Full guide: DEPLOY.md ·
Overview: README

Quick version (using the attached image; the community MCP server must already be running):

# 1. Load the prebuilt image
docker load -i xco-mcp-client-community-1.0.0.tar.gz

# 2. Run it, pointing MCP_BASE_URL at your community MCP server.
#    --add-host lets the container reach a server published on the host.
docker run -d --name xco-mcp-client -p 5174:5174 --restart unless-stopped \
  --add-host=host.docker.internal:host-gateway \
  -e MCP_BASE_URL=http://host.docker.internal:8000 \
  -v xco-mcp-audit:/app/data \
  xco-mcp-client-community:1.0.0

# 3. Open the UI
curl http://localhost:5174/api/health     # {"status":"ok", ...}
# then browse to http://localhost:5174/

⚠️ No login / no authorization — run on localhost or a trusted network
only. The image contains no credentials. Full deploy (reverse proxy,
persistence, networking) is in
DEPLOY.md.

Tested with

Item Version
Client image xco-mcp-client-community:1.0.0
MCP server v1.0.0
Python (container) 3.11.15
Node 20.20.0
Docker 29.3.1
Test date 2026-06-27

Test evidence

Check Result
Backend pytest PASS (148 tests)
Frontend build (npm run build) PASS
Frontend tests (npm test) PASS (98 tests)
Docker build PASS
/api/health PASS
/api/tools (non-empty) PASS (271 tools)
Integration smoke (tests.smoke) PASS (8/8)

Caveats

  • No login / no authorization layer
  • Do not expose directly to the public internet
  • Anyone with access to the UI can invoke read-only MCP tools through the
    configured server
  • Optional OpenAI integration may incur API usage
  • Community / demo support only — no Extreme GTAC