Skip to main content
Glama
deenesjakoruzh
Steph-ux

cloudflare-mcp

by Steph-ux
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Remote

cloudflare-mcp v1.0.0

Universeller Model Context Protocol (MCP)-Server für das gesamte Cloudflare-Ökosystem. Ersetzt 16 fragmentierte Remote-MCP-Endpunkte durch einen einzigen, leistungsstarken lokalen Server.

Funktionen (7 einheitliche Tools)

  • cf_dns: Zonen, DNS-Einträge (A, CNAME, TXT, MX usw.), CDN-Cache-Bereinigung

  • cf_compute: Cloudflare Workers, KV-Key-Value-Speicher, serverlose D1-SQL-Datenbankausführung

  • cf_storage: R2-Objektspeicher-Buckets, Vectorize-Vektordatenbankabfragen

  • cf_ai: Workers-KI-Inferenz (Llama, DeepSeek, Embeddings), AI-Gateway-Protokolle und Telemetrie

  • cf_security: Zero Trust Tunnels (cloudflared), WAF-Firewall-Regeln, Audit-Protokolle

  • cf_radar: Radar-URL-Scanner, globale Verkehrs-/Angriffsanalysen, GraphQL-Abfragen

  • cf_api: Universelles REST-API-Gateway (direkter Zugriff auf alle 2.500+ Cloudflare-Endpunkte)

Related MCP server: mcp-server-cloudflare

Konfiguration

Legen Sie Ihr Cloudflare-API-Token in Ihrer Umgebung fest oder speichern Sie es in ~/.secrets/cloudflare-token.txt:

export CLOUDFLARE_API_TOKEN="your-cloudflare-api-token"
export CLOUDFLARE_ACCOUNT_ID="your-account-id"  # Optional, automatically discovered if omitted

Installation in MCP-Clients (Claude Desktop, OpenCode, Cursor, Windsurf)

Option 1: Sofort über uvx (Empfohlen)

{
  "mcpServers": {
    "cloudflare": {
      "command": "uvx",
      "args": ["--from", "git+https://github.com/Steph-ux/cloudflare-mcp.git", "cloudflare-mcp"]
    }
  }
}

Option 2: Über pip install

pip install git+https://github.com/Steph-ux/cloudflare-mcp.git

Dann in Ihrer MCP-Konfiguration:

{
  "mcpServers": {
    "cloudflare": {
      "command": "cloudflare-mcp"
    }
  }
}

Option 3: Lokaler Klon (Entwicklung)

{
  "mcpServers": {
    "cloudflare": {
      "command": "python",
      "args": ["D:\\Steph\\script\\cloudflare-mcp\\server.py"]
    }
  }
}

Verwendungsbeispiele

1. DNS-Verwaltung

# List zones:
cf_dns(action='list_zones')

# Create DNS record:
cf_dns(
    action='create_record',
    zone_id='example.com',
    type='A',
    name='api',
    content='192.168.10.1',
    proxied=True
)

# Purge CDN cache:
cf_dns(action='purge_cache', zone_id='example.com', purge_everything=True)

2. D1-SQL-Datenbank & Workers

# Query D1 SQL database:
cf_compute(
    action='d1_query',
    database_id='YOUR_D1_DATABASE_ID',
    sql='SELECT * FROM users WHERE active = 1 LIMIT 10;'
)

# Store key-value in KV:
cf_compute(
    action='kv_put',
    namespace_id='YOUR_KV_NAMESPACE_ID',
    key_name='session_token',
    value='xyz123'
)

3. R2-Objektspeicher

# List buckets:
cf_storage(action='list_r2')

# Create bucket:
cf_storage(action='create_r2', bucket_name='my-app-assets', location_hint='weur')

4. Workers AI-Inferenz

# Run Llama 3 on Cloudflare edge:
cf_ai(
    action='run_model',
    model='@cf/meta/llama-3.1-8b-instruct',
    prompt='Write a python snippet for async retry logic.'
)

5. Radar-URL-Scan & Sicherheit

# Scan URL for malware & phishing via Cloudflare Radar:
cf_radar(action='scan_url', url='https://target-to-scan.com')

# List active Zero Trust Tunnels:
cf_security(action='list_tunnels')

6. Universelles REST-Gateway (cf_api)

# Call any specific Cloudflare v4 REST endpoint:
cf_api(method='GET', path='/user')

Testen

pytest tests

Lizenz

MIT

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivitySlowing
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers