The Agent ExaminerIndependent Review Authority
Development

pty-mcp

by io.github.raychao-oao · v0.2.3

Interactive PTY sessions over MCP: local shells, SSH, serial ports, and persistent remote sessions.

GitHub stars
11
Transport
stdio (local process)
Auth
None required
Test status
untested

Install & run

Run the server locally with the package command below.

terminal
$https://github.com/raychao-oao/pty-mcp/releases/download/v0.2.3/pty-mcp-darwin-arm64
mcpbhttps://github.com/raychao-oao/pty-mcp/releases/download/v0.2.3/pty-mcp-darwin-arm64· stdio
mcpbhttps://github.com/raychao-oao/pty-mcp/releases/download/v0.2.3/pty-mcp-darwin-amd64· stdio
mcpbhttps://github.com/raychao-oao/pty-mcp/releases/download/v0.2.3/pty-mcp-linux-amd64· stdio
mcpbhttps://github.com/raychao-oao/pty-mcp/releases/download/v0.2.3/pty-mcp-linux-arm64· stdio

Connect this MCP server to your AI agent

MCP servers plug into any MCP-compatible client (Claude Desktop, Cursor, your own agent runtime, and more). The steps are the same everywhere:

  1. 1. Make sure the runtime is available. This server runs as a local process, so its package runner (e.g. npx / uvx / docker) must be installed on the machine running your agent.
  2. 2. Add it to your MCP client config. Point your client at the command above. Most clients accept an MCP servers map keyed by a name of your choosing.
  3. 3. No credentials needed. This server declares no required secrets — it should work out of the box.
  4. 4. Restart your client and verify.The server’s tools should appear in your agent’s tool list, ready to call.

Want this running without wiring configs yourself? Run this with Alfe →

Repository →
Developmentmcpb

Data from the official MCP registry · last updated 2026-03-24 · ★ 11 stars via GitHub.