What is the LocalSend MCP skill?

It is a specialized capability for AI agents that allows them to send and receive files peer-to-peer using the LocalSend protocol, optimized for Tailscale networks.

Does this skill work behind firewalls?

Yes, it is designed to prioritize Tailscale mesh networking, which effectively bypasses traditional firewall restrictions to ensure reliable connectivity between peers.

Can I use this with the standard LocalSend desktop or mobile app?

Yes, this skill is fully compatible with the standard LocalSend application ecosystem for both discovery and file exchange.

How do agents discover each other using this skill?

Discovery is handled through multiple redundant channels including Tailscale APIs, UDP multicast, NATS pub/sub, and even audible voice announcements.

Is the file transfer process secure?

Transfers are secured at the network level via Tailscale's WireGuard encryption, providing a private, encrypted mesh for all data exchange between agents.

LocalSend P2P File Transfer & Discovery

Name: LocalSend P2P File Transfer & Discovery
Author: plurigrid

byplurigrid

•

协作工具

Enables secure peer-to-peer file transfers and agent discovery across Tailscale networks with redundant connection methods and voice announcements.

This skill integrates the LocalSend protocol with AI agents to facilitate seamless, secure file exchanges and mutual discovery in distributed environments. By leveraging Tailscale for mesh networking, NATS for side-channel communication, and text-to-speech for status announcements, it creates a robust ecosystem for multi-agent collaboration. It features deterministic identity mapping and five levels of connection redundancy to ensure reliable data transfer even behind complex firewalls, making it an essential tool for decentralized development workflows and cognitive augmentation across agent meshes.

主要功能

01Multi-lingual voice synthesis for presence and status announcements

027 GitHub stars

03Deterministic peer identification with color-coding and hostname hashing

04Redundant connection methods including human-readable Tailscale DNS

05Mutual discovery protocol via NATS, UDP multicast, and Tailscale APIs

06Secure P2P file transfers over Tailscale mesh networks and LAN

使用场景

01Establishing a secure, ad-hoc file sharing network between local and remote machines

02Transferring large datasets or code artifacts between distributed AI agents

03Building a multi-agent mesh where agents autonomously announce and discover capabilities

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi skill

For use in Claude.ai and ChatGPT

Download Skill