Cequence MCP CLI

A command-line tool by Cequence Security for easily setting up Cequence MCP (Model Context Protocol) servers with AI clients like Cursor IDE, Claude AI, and Windsurf IDE.

Features

• 🚀 One-command setup for Cursor IDE, Claude AI, and Windsurf IDE
• 🔐 Support for API key authentication
• 🛠️ Cross-platform support (macOS, Windows, Linux)
• ✨ Automatic configuration file management
• 📝 Clear setup instructions and feedback
• 🔒 Powered by Cequence's API security platform
• 📦 introspect an internally-hosted MCP server into a Cequence catalog spec

Installation

You can run the CLI directly using npx without installing it globally:

npx @cequenceai/mcp-cli@latest --help

Or install it globally:

npm install -g @cequenceai/mcp-cli

Usage

Setup with Cursor IDE

npx @cequenceai/mcp-cli@latest cursor --url "https://your-cequence-gateway.com/mcp" --name "Cequence API Gateway"

With API key authentication:

npx @cequenceai/mcp-cli@latest cursor --url "https://your-cequence-gateway.com/mcp" --name "Cequence API Gateway" --api-key "your-api-key"

Setup with Claude AI

npx @cequenceai/mcp-cli@latest claude --url "https://your-cequence-gateway.com/mcp" --name "Cequence API Gateway"

With API key authentication:

npx @cequenceai/mcp-cli@latest claude --url "https://your-cequence-gateway.com/mcp" --name "Cequence API Gateway" --api-key "your-api-key"

Setup with Windsurf IDE

npx @cequenceai/mcp-cli@latest windsurf --url "https://your-cequence-gateway.com/mcp" --name "Cequence API Gateway"

With API key authentication:

npx @cequenceai/mcp-cli@latest windsurf --url "https://your-cequence-gateway.com/mcp" --name "Cequence API Gateway" --api-key "your-api-key"

List configured servers

npx @cequenceai/mcp-cli@latest list

Onboard an internally-hosted MCP server (introspect)

When an MCP server is hosted on a private/internal network, the Cequence control plane can't reach it to auto-discover its tools — which is why such servers often end up registered with 0 tools. The introspect command solves this by running on a machine that can reach the server (e.g. inside your network): it acts as an MCP client, performs the MCP handshake, lists every tool, and writes a bundle you upload in the Cequence AI Gateway UI to register the server with its tools intact.

# No auth
npx @cequenceai/mcp-cli@latest introspect \
  --url "https://internal-mcp.corp.local/mcp" \
  --name "Internal Widgets" --primary-category "Developer Tools"

# Bearer token
npx @cequenceai/mcp-cli@latest introspect \
  --url "https://internal-mcp.corp.local/mcp" --auth bearer --token "$TOKEN"

# API key in a custom header
npx @cequenceai/mcp-cli@latest introspect \
  --url "https://internal-mcp.corp.local/mcp" \
  --auth api-key --api-key "$KEY" --header-name "X-API-Key"

# OAuth 2.0 (opens a browser for the authorization-code flow)
npx @cequenceai/mcp-cli@latest introspect \
  --url "https://internal-mcp.corp.local/mcp" --auth oauth --scope "read"

# SSE transport + extra headers
npx @cequenceai/mcp-cli@latest introspect \
  --url "https://internal-mcp.corp.local/sse" --transport sse \
  --header "X-Tenant: acme" --header "X-Env: prod"

By default this writes a single archive <out-dir>/<slug>.tar.gz — nothing else. Output goes to the current directory unless you pass --out; in locked-down environments where the working directory isn't writable, point it somewhere you can write (e.g. --out ~/mcp-specs).

• <slug>.tar.gz — upload this archive in the Cequence AI Gateway UI (add a remote MCP server → "Upload a spec (CLI)"). Contains create-spec.json plus informational tools.json/manifest.json.
• Use --format dir instead to write the three files loose into <out-dir>/<slug>/ (no archive) — handy for inspecting or diffing.

To register the server, open the Cequence AI Gateway UI, choose to add a remote MCP server, pick the CLI option, and upload the generated .tar.gz.

Note: introspect removes the tool-discovery blocker only. For cloud-hosted Cequence gateways, the dataplane must still be able to reach the internal endpoint at runtime (network route / IP allow-list). For private/hybrid deployments where the dataplane runs inside your network, runtime reachability already exists.

introspect options

• --url, -u: Required — remote MCP server URL
• --transport, -t: streamable-http (default) or sse
• --auth, -a: none (default), bearer, api-key, basic, or oauth
• --token: bearer token (with --auth bearer)
• --api-key / --header-name: API key and the header to carry it (default Authorization)
• --basic-user / --basic-pass: credentials (with --auth basic)
• --scope / --callback-port: OAuth scope(s) and local redirect port (default 8765)
• --header: extra "Name: Value" header, repeatable
• --name, -n / --description, -d: catalog display name & description
• --primary-category / --family / --secondary: catalog categorization
• --out, -o: directory to write output into (default: current directory)
• --format, -f: tgz (default — single .tar.gz bundle) or dir (loose files)
• --timeout: per-request timeout in seconds

Command Options

Common Options

• --url, -u: Required - Cequence MCP server URL
• --name, -n: Server name (defaults to "Cequence MCP Gateway")
• --api-key, -k: API key for authentication (optional)

Examples

1. Basic setup for Cursor:

   npx cequence-mcp-cli@latest cursor -u "https://api.example.com/mcp" -n "Example API via Cequence"

2. Setup with authentication:

   npx cequence-mcp-cli@latest cursor -u "https://api.example.com/mcp" -n "Example API via Cequence" -k "sk-1234567890"

3. Setup for Claude AI:

   npx cequence-mcp-cli@latest claude -u "https://api.example.com/mcp" -n "Example API via Cequence"

4. Setup for Windsurf IDE:

   npx cequence-mcp-cli@latest windsurf -u "https://api.example.com/mcp" -n "Example API via Cequence"

Configuration Locations

The CLI automatically detects your platform and configures the appropriate files:

Cursor IDE

• All platforms: ~/.cursor/mcp.json

Claude AI

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json
• Linux: ~/.config/claude/claude_desktop_config.json

Windsurf IDE

• All platforms: ~/.windsurf/mcp.json

Post-Setup Steps

After configuring Cursor IDE:

1. Restart Cursor IDE
2. Open the command palette (Cmd/Ctrl + Shift + P)
3. Look for MCP-related commands
4. Your Cequence-secured APIs are now available as MCP tools

After configuring Claude AI:

1. Restart Claude AI application
2. The Cequence MCP server should be available in your conversation
3. Start using your secured APIs through Claude's interface

After configuring Windsurf IDE:

1. Restart Windsurf IDE
2. Open the command palette (Cmd/Ctrl + Shift + P)
3. Look for MCP-related commands
4. Your Cequence-secured APIs are now available as MCP tools

About Cequence Security

Cequence Security protects APIs from attacks, abuse, and data theft. This CLI tool enables you to easily connect your Cequence-secured APIs to AI development environments, providing secure and monitored access to your API resources.

Troubleshooting

Common Issues

1. Permission errors: Make sure you have write permissions to the configuration directories
2. Invalid URL: Ensure your Cequence MCP server URL is accessible and uses HTTP/HTTPS
3. Configuration not applied: Try restarting the AI client application

Getting Help

npx @cequenceai/mcp-cli@latest --help
npx @cequenceai/mcp-cli@latest cursor --help
npx @cequenceai/mcp-cli@latest claude --help
npx @cequenceai/mcp-cli@latest windsurf --help

Development

To contribute to this CLI tool:

1. Clone the repository
2. Install dependencies: npm install
3. Build: npm run build
4. Test locally: npm run dev

License

MIT License - see LICENSE file for details.

Support

For support with Cequence products or this CLI tool, please visit Cequence Security Support or open an issue in this repository.