Server facts
aiterm-mcp
Project
- Author
- kitepon-rgb
- Repository
- github.com/kitepon-rgb/aiterm-mcp
- License
- Not specified
- Category
- Developer Tools
Tech
- Runtime / language
- Not specified
- Transports
- stdio
- Remote available
- No
- Authentication
- Not specified
Activity
- GitHub stars
- 6
- Last commit
- Data as of
Tags
- Claude
- Terminal
- cli
- devtools
- mcp
- pty
- ssh
- tmux
Overview
SSH sessions, container shells, and REPLs run inside aiterm-mcp’s persistent tmux terminal rather than separate connections for each command. AI agents using stdio MCP clients can open a terminal, send text or control keys, inspect active sessions, and close them through the PTY tools. Sessions survive server or client restarts, and a person can work alongside the agent through the shared tmux socket. Output reads default to incremental updates and reduce terminal noise through control-character removal, repetition compression, and head-and-tail folding.
Dedicated launchers bring Claude Code, Codex, Grok, and Composer into persistent interactive sessions for delegated work. Follow-up input uses pty_send, with agent requests dispatched without waiting for completion; a separate wait process provides completion notification. Transcript reads retrieve the final assistant response from the most recently completed turn when terminal output is truncated. Additional controls change an agent’s model or reasoning effort without restarting its conversation. Claude-specific tools expose structured turn recovery and approval handling tied to the active operation, while read-only diagnostics return status information without terminal contents, credentials, paths, or environment values.
aiterm-mcp setup notes
- aiterm-mcp records stdio only; confirm a network transport upstream before planning remote access.
- aiterm-mcp has no recorded license; confirm reuse terms in its repository.