Discover MCPs & agents
Loading MCPs and agents…
Loading MCPs and agents…
Experimental command-line interface for interacting with the Model Context Protocol (MCP) Registry Service.
From the repo.
A command-line interface for interacting with the MCP Registry Service and connecting directly to MCP servers.
Registry Service Integration:
Direct MCP Server Connection:
Clone the repository and build using Make:
git clone https://github.com/jbovet/mcp-cli.git
cd mcp-cli
make build
The binary will be created in bin/mcp-cli.
# List all servers
mcp-cli get servers
# Get server details by ID
mcp-cli get server 123e4567-e89b-12d3-a456-426614174000
# Get server details by name
mcp-cli get server io.github.owner/server-name
# Search servers by name pattern
mcp-cli get server redis --name --show-matches
# Check service health
mcp-cli health
# Ping service
mcp-cli ping
# Connect to a Python MCP server
mcp-cli connect --type stdio --command "python" --args "server.py"
# Connect to a Node.js MCP server with environment variables
mcp-cli connect --type stdio --command "node" --args "server.js" --env "DEBUG=1"
# Connect using command string parsing
mcp-cli connect --type stdio --command "python /path/to/server.py --port 8080"
# Interactive mode for exploration
mcp-cli connect --type stdio --command "python server.py" --interactive
# Connect to HTTP MCP server
mcp-cli connect --type http --url "http://localhost:8080/mcp"
# Connect to HTTPS MCP server
mcp-cli connect --type streamable --url "https://api.example.com/mcp"
# Interactive mode with HTTP server
mcp-cli connect --type http --url "http://localhost:8080/mcp" --interactive
When running with --interactive, you can use these commands:
# List server capabilities
tools # List available tools
resources # List available resources
prompts # List available prompts
# Execute operations
call <tool-name> [args...] # Call a tool with arguments
read <resource-uri> # Read a resource by URI
# Navigation
help # Show help message
quit/exit # Exit interactive mode
> tools
Available tools (3):
1. validate_api - Validate API specifications
2. format_code - Format source code
3. analyze_logs - Analyze log files
> call validate_api openapi.yaml
Tool 'validate_api' result:
✅ Tool executed successfully:
API specification is valid
Found 12 endpoints
No validation errors
> resources
Available resources (2):
1. file://config.json (Configuration) - Server configuration
2. file://schema.yaml (Schema) - API schema definition
> read file://config.json
Resource content:
Text (application/json): {"server": {"port": 8080}}
> quit
Goodbye!
--url: Base URL of the MCP Registry Service (default: http://localhost:8080)--verbose, -v: Enable verbose output--limit: Maximum number of servers to return (1-100)--cursor: Pagination cursor for fetching next page--type: Transport type (stdio, http, streamable)--url: Server URL for HTTP-based connections--command: Command to execute for stdio connections--args: Arguments for the command (can be repeated)--env: Environment variables for the command (can be repeated)--timeout: Connection timeout (default: 60s)--interactive: Run in interactive mode# Install dependencies
make deps
# Run tests
make test
# Format code
make fmt
# Run linter
make lint
# Clean build artifacts
make clean
cmd/ - Command implementations
connect.go - MCP server connection command
get.go - Registry "get" command group
server.go - Individual server details
servers.go - Server listing
health.go - Health check command
ping.go - Ping command
pkg/ - Core packages
client/ - Registry API client implementation
models/ - Data models
adapter/ - MCP server adapters (stdio, HTTP)
adapter.go - Core adapter interfaces
stdio.go - Stdio transport implementation
http.go - HTTP transport implementation
factory.go - Adapter factory and utilities
bin/ - Build output
MIT License - see LICENSE.md for details.
Replace {MCP_ENDPOINT_URL} with this MCP’s endpoint URL (from its repo or docs above). No API key — you connect directly.
Tool
OS
Config file: ~/.cursor/mcp.json
{
"mcpServers": {
"mcp-server": {
"url": "{MCP_ENDPOINT_URL}"
}
}
}Paste into mcpServers in the config file. Restart Cursor after saving.
If this MCP is also published on mcpchannel.ai, you can subscribe from Browse and use the gateway config there instead.