Download

Native app for macOS, Windows, and Linux.

Latest release: v0.0.6

macOS

Universal (Apple Silicon & Intel)

.app

Windows

x86_64

.exe

Linux

x86_64

binary

CLI Install

curl -fsSL https://latere.ai/wallfacer/install.sh | sh

Detects your OS and architecture, downloads the latest binary to /usr/local/bin (or ~/.local/bin). Override with WALLFACER_INSTALL_DIR or pin a version with WALLFACER_VERSION=v0.0.6.

Prerequisites

Podman or Docker — Wallfacer auto-detects whichever is available. Required for task execution in isolated containers.
A Claude credential — either a Claude Pro/Max OAuth token or an Anthropic API key (configured after install in Settings).
Git (recommended) — non-git directories work as workspaces but git features (worktrees, diff, auto-push) are unavailable.

Quick Start

1. Check prerequisites

wallfacer doctor

Checks configuration paths, credentials, container runtime, sandbox images, and Git. Items marked [ok] are ready, [!] need attention, [ ] are optional.

2. Start the server

wallfacer run

Opens browser to http://localhost:8080. The sandbox image (ghcr.io/changkun/wallfacer:latest, ~1 GB) is pulled automatically on first task run.

3. Add your credential

Open Settings in the browser. Choose one:

Sign in with Claude (easiest) — click the button, authenticate in browser
OAuth token — paste into CLAUDE_CODE_OAUTH_TOKEN (obtain via claude setup-token)
API key — paste sk-ant-... key into ANTHROPIC_API_KEY

4. Create your first task

Click + New Task, enter a prompt, and drag the card to In Progress. A sandbox container starts and live logs appear in the detail panel.

CLI Reference

wallfacer run [flags]              # Start the server
wallfacer doctor                   # Check prerequisites
wallfacer status                   # Print board state
wallfacer status -watch            # Live-updating board
wallfacer status -json             # JSON output
wallfacer exec <task-id>           # Attach to container
wallfacer exec --sandbox claude    # New sandbox shell

Common flags for wallfacer run:

`-addr`	`:8080`	Listen address
`-no-browser`	`false`	Skip auto-opening browser
`-container`	auto	Container runtime (`podman` or `docker`)
`-log-format`	`text`	Log format: `text` or `json`