Backend

A Backend is a VT-emulator implementation: the engine that takes a stream of bytes (text and ANSI escape sequences) and maintains a terminal buffer — cells, colors, cursor, scrollback, modes.

A Backend is the live side's foundation. It does one job: parse bytes into state. It does not own a PTY, does not spawn processes, does not render pixels — a Terminal wraps a Backend to add those.

Why termless has many backends

Real terminal emulators disagree. xterm.js and Ghostty render the same emoji at different widths; different emulators encode the same keypress differently, clamp colors differently, and scroll differently. Termless ships multiple backends so you can run the same test or recording against each one and find exactly where they diverge.

Backend	Engine	Notes
xtermjs	@xterm/headless	Fast, portable — the default
ghostty	ghostty-web (WASM)	Truecolor + full glyph coverage; visual work
vt100	pure TypeScript	Minimal VT100-era subset, zero native deps
vterm	pure TypeScript	Full standards coverage
alacritty	alacritty_terminal (napi-rs)	Needs a Rust build
wezterm	wezterm-term (napi-rs)	Needs a Rust build
peekaboo	OS automation against a real app	macOS only — pixel-perfect, slowest

See the Backends guide for the full capability matrix and the Backend API for the interface every backend implements.

Choosing a backend

Two ways to pick one:

// Factory function — explicit, synchronous
import { createXtermBackend } from "@termless/xtermjs"
import { createTerminal } from "@termless/core"
const term = createTerminal({ backend: createXtermBackend() })

// By name — async, handles WASM/native init
import { backend } from "@termless/core"
const term = createTerminal({ backend: await backend("ghostty") })

Default to xtermjs for speed; reach for ghostty when visual fidelity matters (screenshots, visual-bug evidence); reach for peekaboo only when you need a real desktop terminal app in the loop.