mcp-cloudflare

Slim Cloudflare MCP Server for managing DNS, zones, tunnels, WAF, Zero Trust, and security via Cloudflare API v4.

No SSH. No shell execution. API-only. 3 runtime dependencies.

Table of Contents

• Features
• Quick Start
• HashiCorp Vault Integration (Optional)
• Claude Code Integration
• Configuration
• Multi-Zone Support
• Tools
• Skills
• Development
• License

Features

75 tools across 11 domains:

• DNS — Record management (A, AAAA, CNAME, MX, TXT, SRV, CAA, NS), batch operations
• Zones — Zone listing, settings, SSL/TLS configuration, cache management
• Tunnels — Cloudflare Tunnel creation, configuration, and ingress management
• WAF — Ruleset management, custom firewall rules, rate limiting
• Zero Trust — Access application CRUD (create/delete), policies (create/delete), identity providers (create/delete), Gateway status
• Security — Security event analytics, IP access rules, DDoS configuration, Security Center insights
• Workers KV — Namespace management, key-value read/write/delete, key listing
• Workers — Script deployment, route management
• Worker Secrets — Secret management (names only, values never exposed)
• Worker Analytics — Invocation metrics, CPU time, error rates via GraphQL
• R2 Storage — Bucket management, object listing and metadata, custom domains, location hints

Quick Start

npm install
cp .env.example .env   # Edit with your Cloudflare API token
npm run build
node dist/index.js     # stdio transport for MCP

HashiCorp Vault Integration (Optional)

mcp-cloudflare supports loading Cloudflare credentials from a central HashiCorp Vault instance at startup via AppRole authentication. This is optional — the server works fine with plain environment variables alone.

How It Works

On startup, if NAS_VAULT_ADDR is set the server performs an AppRole login, fetches the KV v2 secret at <mount>/data/cloudflare/api, and injects the values into the process environment before the MCP transport starts. The loader is fully opportunistic:

• If NAS_VAULT_ADDR is unset, the loader is a silent no-op. No Vault calls are made and the server behaves exactly as before.
• On any Vault error (network failure, bad credentials, missing secret path), a single-line warning is written to stderr and the server falls back to whatever environment variables are already set.
• Secret values are never logged. Only the KV path name and a populated-count appear in stderr diagnostics.
• Uses the built-in fetch (Node 20+) — no additional runtime dependencies.

Credential Precedence

Explicit env vars (CLOUDFLARE_API_TOKEN etc.) > Vault > error (missing creds)

If you set CLOUDFLARE_API_TOKEN directly, the Vault loader will not overwrite it. Vault only fills in credentials that are not already present in the environment.

Vault Environment Variables

* Only required if using Vault. All three must be set together.

KV v2 Secret Structure

Write the Cloudflare credentials to the following path in Vault:

Path: kv/cloudflare/api

{
  "api_token": "your-cloudflare-api-token",
  "account_id": "your-account-id"
}

Key mapping:

Vault Setup Steps

1. Write credentials to KV v2:

vault kv put kv/cloudflare/api \
  api_token="your-cloudflare-api-token" \
  account_id="your-account-id"

2. Create a Vault policy:

# cloudflare-mcp-policy.hcl
path "kv/data/cloudflare/api" {
  capabilities = ["read"]
}

vault policy write cloudflare-mcp cloudflare-mcp-policy.hcl

3. Enable AppRole auth and create a role:

vault auth enable approle

vault write auth/approle/role/cloudflare-mcp \
  token_policies="cloudflare-mcp" \
  token_ttl="1h" \
  token_max_ttl="4h" \
  secret_id_ttl="0"   # 0 = no expiry; set a duration for rotation

4. Retrieve the role ID and secret ID:

vault read auth/approle/role/cloudflare-mcp/role-id
vault write -f auth/approle/role/cloudflare-mcp/secret-id

Claude Desktop / MCP Config Example (with Vault)

When using Vault, no Cloudflare credentials are needed in the MCP config — only the three Vault variables:

{
  "mcpServers": {
    "cloudflare": {
      "command": "npx",
      "args": ["@itunified.io/mcp-cloudflare"],
      "env": {
        "NAS_VAULT_ADDR": "https://vault.example.com:8200",
        "NAS_VAULT_ROLE_ID": "your-role-id",
        "NAS_VAULT_SECRET_ID": "your-secret-id"
      }
    }
  }
}

NAS_VAULT_KV_MOUNT can be omitted if your KV engine is mounted at the default path kv. The Cloudflare API token and account ID will be fetched automatically at startup.

Claude Code Integration

Add to .mcp.json in your project root:

{
  "mcpServers": {
    "cloudflare": {
      "command": "node",
      "args": ["/path/to/mcp-cloudflare/dist/index.js"],
      "env": {
        "CLOUDFLARE_API_TOKEN": "your-api-token-here",
        "CLOUDFLARE_ACCOUNT_ID": "your-account-id"
      }
    }
  }
}

Configuration

Loading Secrets from HashiCorp Vault (AppRole)

If you run a central Vault instance, mcp-cloudflare can fetch its credentials at startup via AppRole instead of passing them through the MCP config:

export NAS_VAULT_ADDR=https://vault.example.com
export NAS_VAULT_ROLE_ID=<role-id>
export NAS_VAULT_SECRET_ID=<secret-id>
# optional — defaults to "kv"
export NAS_VAULT_KV_MOUNT=kv

The loader reads KV v2 at <mount>/data/cloudflare/api and expects two keys: api_token and account_id. Example Vault write:

vault kv put kv/cloudflare/api \
  api_token=your-api-token-here \
  account_id=00000000000000000000000000000000

Precedence: process.env (explicit) > Vault. If NAS_VAULT_ADDR is unset the loader is a silent no-op — the server behaves exactly as before. On any Vault error (network, auth, missing path), a single-line warning is written to stderr and the server falls back to whatever env vars are already set.

Security: secret values are never logged. Only the KV path name and a populated-count appear in stderr diagnostics. Uses the global fetch (Node 20+) — no new runtime dependencies.

API Token Permissions

Create an API Token at dash.cloudflare.com/profile/api-tokens with the following permissions based on what you need:

• DNS: Zone > DNS > Edit
• Zone settings: Zone > Zone Settings > Edit
• Cache purge: Zone > Cache Purge > Edit
• Tunnels: Account > Cloudflare Tunnel > Edit
• WAF: Zone > Firewall Services > Edit
• Zero Trust: Account > Access: Apps and Policies > Edit
• Security events: Zone > Analytics > Read
• Workers KV: Account > Workers KV Storage > Edit
• Workers: Account > Worker Scripts > Edit
• R2: Account > R2 Storage > Edit

Multi-Zone Support

All zone-scoped tools accept a zone_id parameter that can be either:

• A 32-character hex zone ID (e.g., 00000000000000000000000000000001) — used directly
• A zone name / domain (e.g., example.com) — resolved automatically via the Cloudflare API

This allows managing multiple zones by name without needing to look up IDs manually.

Tools

Tools documentation is coming in v1 as tool modules are implemented. See docs/api-reference.md for the planned API endpoint mapping.

Skills

Claude Code skills compose MCP tools into higher-level workflows. See .claude/skills/README.md for detailed documentation.

Development

npm run build      # Compile TypeScript
npm test           # Run unit tests (vitest)
npm run typecheck  # Type check only (no emit)

See CONTRIBUTING.md for contribution guidelines.

License

This project is dual-licensed:

• Open Source: GNU Affero General Public License v3.0 (AGPL-3.0) — free for open-source and non-commercial use
• Commercial: Available for proprietary integrations — see COMMERCIAL_LICENSE.md

If you use mcp-cloudflare in a proprietary product or SaaS offering, a commercial license is required. Support development by sponsoring us on GitHub.