NAME
Net::Async::MCP::Server - Async MCP server for command execution
VERSION
version 0.001
SYNOPSIS
use IO::Async::Loop;
use Net::Async::MCP::Server;
my $loop = IO::Async::Loop->new;
my $server = Net::Async::MCP::Server->new(
name => 'command-runner',
allowed_commands => ['ls', 'cat', 'grep'],
working_directory => '/tmp',
timeout => 30,
compress => 1,
);
$loop->add($server);
async sub main {
await $server->initialize;
my $tools = await $server->list_tools;
# [{name => 'run', description => '...', inputSchema => {...}}]
my $result = await $server->call_tool('run', {
command => 'ls -la',
working_directory => '/tmp',
timeout => 10,
});
# {content => [{type => 'text', text => '...'}], isError => 0}
}
main()->get;DESCRIPTION
Net::Async::MCP::Server is an asynchronous MCP (Model Context Protocol) server for command execution, built on IO::Async and Future::AsyncAwait.
Extends Net::Async::MCP to act as an MCP server (not client) that exposes a run tool for command execution. The async counterpart to MCP::Run::Bash.
ATTRIBUTES
name (required)
The server name exposed via MCP protocol.
allowed_commands
ArrayRef of command names (first words) that are permitted. Commands not in this list are rejected with an error result.
$server->allowed_commands(['ls', 'cat', 'grep']);working_directory
Default working directory for command execution. Can be overridden per-invocation.
$server->working_directory('/var/data');timeout
Default timeout in seconds. Defaults to 30.
$server->timeout(60);compress
Enable output compression via MCP::Run::Compress. When enabled, command output is filtered through compression filters before returning to the client.
$server->compress(1);validator
Coderef that validates a command before execution.
$server->validator(sub ($cmd, $dir) {
return 1 if $cmd =~ /^ls|^cat|^grep/;
return "blocked by policy";
});METHODS
async method initialize()
Performs MCP server initialization. Handles the initialize request from the client, sends server info and capabilities, registers the 'run' tool.
async method list_tools()
Returns ArrayRef of tool definitions. Returns the registered 'run' tool.
async method call_tool($name, $arguments)
Calls the specified tool with arguments. For the 'run' tool:
my $result = await $server->call_tool('run', {
command => 'ls -la',
working_directory => '/tmp',
timeout => 10,
});run_stdio class method
Convenience constructor that creates a server and runs it in stdio mode.
Net::Async::MCP::Server->run_stdio(name => 'my-server');method handle($request, $context)
Handles a JSON-RPC request and returns a response. This is the main entry point for the stdio transport.
my $response = $server->handle($request);Returns undef for notifications (no response needed).
SEE ALSO
Net::Async::MCP, MCP, MCP::Server, MCP::Run, MCP::Run::Bash.
SUPPORT
Issues
Please report bugs and feature requests on GitHub at https://github.com/Getty/p5-net-async-mcp-run/issues.
CONTRIBUTING
Contributions are welcome! Please fork the repository and submit a pull request.
AUTHOR
Torsten Raudssus <torsten@raudssus.de>
COPYRIGHT AND LICENSE
This software is copyright (c) 2026 by Torsten Raudssus <torsten@raudssus.de> https://raudssus.de/.
This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.