| name | claude-term |
| description | Use when you need persistent terminals — long-running processes, dev servers, REPLs, or any command you want to check back on later. |
claude-term — Persistent Terminals
Use claude-term when you need a terminal that persists beyond a single command — long-running processes, interactive tools, or terminals you want to check back on later.
When to Use
- Use claude-term for: servers, watchers, builds, REPLs, SSH sessions, anything long-running or interactive
- Use the Bash tool for: quick one-shot commands (ls, git status, grep)
Quick Reference
claude-term spawn --cwd /path/to/project
claude-term spawn bash -c "npm run dev"
claude-term list
claude-term write t1 "npm test\n"
claude-term read t1
claude-term kill t1
Workflow Example
TERM_ID=$(claude-term spawn --cwd ~/projects/myapp bash -c "npm run dev")
sleep 2
claude-term read "$TERM_ID"
TEST_ID=$(claude-term spawn --cwd ~/projects/myapp bash -c "npm test")
sleep 5
claude-term read "$TEST_ID"
claude-term kill "$TERM_ID"
claude-term kill "$TEST_ID"
Owner Tracking
Terminal ownership is automatic. When you spawn a terminal, it's automatically associated with your session. claude-term list shows only your terminals by default. Use claude-term list --all to see all terminals.
Terminal Lifecycle
Terminals persist until explicitly killed or inactive for 24 hours (no reads, no writes, no output). They survive session disconnects and restarts.
Interactive Attach
For interactive use (humans, not agents), claude-term attach <id> provides a live bidirectional connection. Press Ctrl+] to detach without killing the terminal.