by bytedance · bytedance/UI-TARS-desktop
ByteDance's open multimodal agent stack — vision-grounded computer-use, callable from Claude via MCP.
UI-TARS Desktop is ByteDance's open multimodal agent platform. It uses screen understanding for desktop control and exposes itself as MCP. Claude can delegate vision-grounded GUI tasks where text-only tools fall short.
%APPDATA%\Claude\claude_desktop_config.json{
"mcpServers": {
"ui-tars-desktop-mcp": {
"command": "docker",
"args": [
"run",
"-p",
"8080:8080",
"ghcr.io/bytedance/ui-tars-desktop:latest"
]
}
}
}Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.
{
"mcpServers": {
"ui-tars-desktop-mcp": {
"command": "docker",
"args": [
"run",
"-p",
"8080:8080",
"ghcr.io/bytedance/ui-tars-desktop:latest"
]
}
}
}Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.
{
"mcpServers": {
"ui-tars-desktop-mcp": {
"command": "docker",
"args": [
"run",
"-p",
"8080:8080",
"ghcr.io/bytedance/ui-tars-desktop:latest"
]
}
}
}Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".
{
"mcpServers": {
"ui-tars-desktop-mcp": {
"command": "docker",
"args": [
"run",
"-p",
"8080:8080",
"ghcr.io/bytedance/ui-tars-desktop:latest"
]
}
}
}Same shape as Claude Desktop. Restart Windsurf to pick up changes.
{
"mcpServers": [
{
"name": "ui-tars-desktop-mcp",
"command": "docker",
"args": [
"run",
"-p",
"8080:8080",
"ghcr.io/bytedance/ui-tars-desktop:latest"
]
}
]
}Continue uses an array of server objects rather than a map.
{
"context_servers": {
"ui-tars-desktop-mcp": {
"command": {
"path": "docker",
"args": [
"run",
"-p",
"8080:8080",
"ghcr.io/bytedance/ui-tars-desktop:latest"
]
}
}
}
}Add to context_servers. Zed hot-reloads on save.
claude mcp add ui-tars-desktop-mcp -- docker run -p 8080:8080 ghcr.io/bytedance/ui-tars-desktop:latestOne-liner. Verify with claude mcp list. Remove with claude mcp remove.
When to use: You're automating a desktop or canvas app that has no scripting surface.
Outcome: GUI automation in apps with no API.
UI-TARS for vision tasks; OpenAgent for code-text tasks
| Tool | Inputs | When to call | Cost |
|---|---|---|---|
| ui_act | (see docs) | Run a vision-grounded GUI action | 1 call |
| ui_observe | (see docs) | Capture screen state | 1 call |
screen-capture input-injectionRun with banking app visibleAction misses target on Retina displaySet the DPI scale; UI-TARS defaults to 1x
| Alternative | When to use it instead | Tradeoff |
|---|---|---|
| OpenAgent | Lower resource cost | Less vision grounding |