YAFL MCP Server
Enables AI agents to securely transfer files between machines via encrypted, expiring share links, with tools for upload, download, status checks, and link management.
README
YAFL — Yet Another File Layer (MCP server + CLI)
Your AI agent moves files between machines with one MCP call — encrypted client-side, links gone in 24 hours.
This repository is the open-source MCP server and CLI client for YAFL. Your agents (Claude Desktop, Claude Code, Cursor, any MCP client — including agents on remote VPSes) hand files across environments — laptop ↔ VPS, agent ↔ agent — with a single tool call, instead of a human dragging files into WeTransfer and copying links around. Files are encrypted on your machine before they leave it; the service only ever sees ciphertext.
The client is open source so you can audit exactly what touches your files. The YAFL service lives at yafl.dev.
Install
Add YAFL to your MCP client config:
{
"mcpServers": {
"yafl": {
"command": "npx",
"args": ["-y", "@yafldev/mcp"],
"env": { "YAFL_API_KEY": "<your key — or leave unset and run yafl login>" }
}
}
}
Works in Claude Desktop, Claude Code, Cursor, and any MCP client. Claude Desktop users can install the one-click .mcpb bundle. Prefer the CLI:
curl -fsSL https://yafl.dev/install.sh | sh
No API key is required to start. On the first tool call the server returns a guided error pointing to yafl login — an RFC 8628 device flow: the agent shows a short code, you approve it in a browser, and the agent holds its own revocable key. Built for headless boxes. Runs via npx (Node.js 18+); no global install needed.
Tools
| Tool | Call | What it does |
|---|---|---|
upload_file |
{ path, password?, oneTime? } |
Encrypts client-side, uploads ciphertext, returns the share link. Zips a directory first. Optional password layer; optional one-time link that burns after the first download. |
download_file |
{ url, destDir?, password? } |
Fetches, decrypts, writes the file to local disk. |
get_status |
{ url } |
Checks liveness and remaining TTL — no bytes downloaded. |
list_files |
{ limit? } |
The caller's own transfers, metadata only. |
delete_file |
{ id | url } |
Revoke one of your transfers before expiry. |
email_link |
{ link, recipient } |
Email a share link straight to a recipient — the agent-to-human handoff without leaving the session. |
login |
device flow | Authorize this agent (RFC 8628); it holds its own revocable key. |
Errors are stable and machine-readable — your agent gets expired or not_found, not a mystery 404.
How encryption works
The file is encrypted with a fresh AES-256-GCM key before upload. The key travels only in the share URL's #fragment — browsers never send fragments to servers, so it is never transmitted to, logged by, or stored on any server. An optional link password adds a second derived layer the server never sees either, and a one-time link self-destructs after the first successful download. File bytes go directly to storage via pre-signed URLs; the app server never handles them, and storage only ever receives ciphertext. Access stops at 24 hours. Free tier: 100 MB per file.
What happens if an agent's key leaks
The API key is not the decryption key. If a machine is compromised and its YAFL key is stolen, the attacker can at most see ciphertext metadata for that key's own transfers — file contents stay unreadable, because decryption keys exist only in the links you exchanged. Revoke the key and move on. This is why the recommended setup is one key per agent, per machine — keys are named and individually revocable.
Links
- Website: https://yafl.dev
- MCP guide: https://yafl.dev/mcp
- Security / threat model: https://yafl.dev/security
- npm:
@yafldev/mcp·@yafldev/cli
License
MIT — see LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.