Lets an AI client drive a Raspberry Pi over an end-to-end encrypted channel, providing tools to run shell commands, read and write files, install apt packages, and fetch pending results or health status through the PNA Vercel API.
Uses the PNA Vercel API as the relay backend that stores opaque encrypted request and response blobs between the AI client and the Raspberry Pi, without ever seeing plaintext; the API base can be overridden via PNASYS_VERCEL_BASE.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@PNASystems CRP MCP Server check the CPU temp on my Pi using api_key abc123 and session_key s-9f2"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Lets an AI client drive a Raspberry Pi through the PNA Vercel API over an
end-to-end encrypted channel: the AI supplies the Pi's api_key and the
session_key printed by pnasyscrp enable. Requests are encrypted locally
with pnasys-encryption-service, Vercel stores the opaque blob, the Pi
decrypts it; the Pi encrypts its response the same way and this server
decrypts it. Vercel never sees plaintext.
pip install pnasys-crp-mcp
# then run from PATH:
pnasys-crp-mcpOr without installing, via uvx:
uvx pnasys-crp-mcpRelated MCP server: MCP Secure Relay
pip install git+https://github.com/Goplop0959/pnasys-crp-mcp.git
pnasys-crp-mcpOr run it without installing, straight from the repo:
uvx --from git+https://github.com/Goplop0959/pnasys-crp-mcp.git pnasys-crp-mcp --versionSame for OpenCode — swap the command to pull from GitHub instead of PyPI:
{
"mcp": {
"pnasys-crp": {
"type": "local",
"command": ["uvx", "--from", "git+https://github.com/Goplop0959/pnasys-crp-mcp.git", "pnasys-crp-mcp", "--AccessKey", "PASTE_ACCESS_KEY_FROM_SETUP"],
"enabled": true
}
}
}opencode.json){
"mcp": {
"pnasys-crp": {
"type": "local",
"command": ["uvx", "pnasys-crp-mcp", "--AccessKey", "PASTE_ACCESS_KEY_FROM_SETUP"],
"enabled": true
}
}
}With a PATH install instead: "command": ["pnasys-crp-mcp", "--AccessKey", "..."].
The access key is printed once by pnasyscrp setup; each pnasyscrp enable
prints a fresh session key, which you hand to the AI per tool call
(session_key param — pass api_key empty to use the baked-in default).
Override the API base with PNASYS_VERCEL_BASE env (default production).
pi_secure_exec(api_key, session_key, cmd, wait_s) — run shell
pi_secure_read(api_key, session_key, path, wait_s) — read file
pi_secure_write(api_key, session_key, path, data_b64, wait_s) — write file
pi_secure_install(api_key, session_key, pkg, wait_s) — apt install
pi_fetch_result(api_key, session_key, request_id) — collect a pending response
pi_health() — API reachability
Each call encrypts, enqueues a secure job, waits for the Pi (long-poll),
and returns the result. Nothing secret is logged; keys stay parameters.
This server cannot be deployed
Use your Mac, Windows or Linux computer from ChatGPT, Claude or Codex: files, commands, documents.
Remote shell and detached long-running jobs on your own machines — no SSH, open ports or VPN.
Drive real devices from your AI Coding tool. Embed a client SDK (Unity, Godot, Flutter, iOS/macOS, Android, React Native, Web) in your app, then capture screenshots, traverse the UI tree, inject taps and key events, and run automated test tasks on the physical device over a secure relay.
Operate Linux, macOS and Windows from your LLM. Every action runs through an auditable allowlist.
