TERMINAL-FIRST · POWERED BY ANYBRIDGE

An AI coding agent,
native to your terminal.

AnyBridge is an AI coding agent that takes on real, multi-step work. Plan a feature, refactor across dozens of files, track down a failing test, ship the fix, all without ever leaving your terminal. Sign in with your AnyBridge account and put it to work.

Install options →

Linux & macOS · then anybridge login with your AnyBridge account.

// what you get

Everything you'd want from a coding agent, on your terms.

01

Tackles complex, multi-step tasks

Hand it a real job: a new feature, a cross-cutting refactor, a failing test. It plans the work, edits across many files, runs your commands and tests, and iterates until it's done, all in a keyboard-driven TUI that feels right over SSH.

02

Moves fast without going rogue

Start in Plan mode to explore read-only and agree on an approach (Shift+Tab to toggle). Then it executes inside the sandbox you choose (read-only, workspace-write, or full access), with approval prompts gated exactly how you like.

03

Fits how you work

Drive it interactively in the TUI, fire one-shot tasks with anybridge exec, get a non-interactive code review with anybridge review, or resume and fork past sessions. You can even run AnyBridge itself as an MCP server.

04

Plugs into your ecosystem

Bring the tools you already use. AnyBridge is compatible with the open ecosystem of MCP servers, skills, and plugins. Add lifecycle hooks, spin up subagents, and let it carry memory across sessions.

// install

One line. Then you're coding.

The installer detects your OS and architecture, verifies the download's SHA-256 checksum, drops the anybridge binary into ~/.anybridge/bin, and adds it to your PATH.

  • Linux x86_64 · arm64
  • macOS x86_64 · arm64
# pin a version
VERSION=v1.2.3 curl -fsSL https://code.anybridge.ai/install.sh | bash

// first commands

Get going in seconds.

anybridgeopen the TUI in the current directory
anybridge exec "fix the failing test"one-shot: run a task and exit
anybridge reviewreview your changes, non-interactively
anybridge resumepick up a previous session
anybridge loginsign in with your AnyBridge account
anybridge doctorcheck install, config & auth health
anybridge updateupdate to the latest release
/modelchoose the model and reasoning effort
/planexplore read-only in Plan mode (or Shift+Tab)
/initgenerate an AGENTS.md for your project
/mcplist and manage MCP tools
/pluginsbrowse and add plugins

On a remote machine over SSH?
Sign in with anybridge login --device-auth (or /login → Device code).

Bring the agent to your machine.