Best for
- Developers who need to invoke commands through MCP
- Local automation scenarios where command permissions and working directories can be managed
U
UI-TARS Desktop MCP Server
Community
An MCP server for running arbitrary commands
- Category
- Dev Tools #380 of 438
- Stars
- ★ 39.3k Very popular
- Transport
- stdio (local process) · SSE · Streamable HTTP
- Runtime
- Node.js 22+
- Credentials
- No credential needed
- License
- Apache-2.0
- Last commit
36FMRS · D
The source lists this server as active at version 1.0.0 and documents three transport options. No tool names, client compatibility, or authentication configuration are provided, so they are not inferred.
Strongest · Maintenance 12/20 Weakest · Security and permissions 2/20
Why each score
- Reliability 7/20
- The manifest identifies an npm package, three transports, and cwd/port parameters, making the launch path appear plausible. However, the supplied material cannot verify MCP initialization, the tool-list handshake, actual tool behavior, error handling, or dependency control. There is no server-specific tool schema, test coverage, or execution evidence, and the arbitrary-command capability adds uncertainty, so points are deducted conservatively.
- Security and permissions 2/20
- The server is explicitly described as running arbitrary commands, a high-privilege capability that may cause deletion, deployment, remote-operation, or other side effects. The material does not establish confirmation prompts, command allowlists, sandboxing, privilege isolation, auditing, or data boundaries. This is treated under the red line for dangerous or remote-command capability without demonstrated confirmation, so security is scored at 2.
- Maintenance 12/20
- The repository is not archived, the manifest marks the server active, the project uses Apache-2.0, and the README reports project releases and updates through 2025, indicating ongoing activity. However, there is no server-specific evidence of commits, tests, dependency updates, a security-response channel, or a pinned release strategy. The use of the mutable npm tag latest also weakens reproducibility, so substantial points are deducted.
- Documentation 6/20
- The repository README documents the broader Agent TARS/UI-TARS projects, MCP integration, and general usage, while the manifest supplies the package identifier, transport endpoints, and cwd/port parameters. It does not document this commands server's tools, parameter constraints, dangerous-operation limits, authentication, cost, error behavior, or troubleshooting. Sibling-server and broader-project material cannot be attributed to this server, resulting in a low score.
- Setup experience 9/20
- The manifest provides an npm registry, package identifier, stdio transport, local SSE and streamable-http endpoints, and an npx runtime hint, which is a usable installation starting point. It lacks a complete configuration example for a mainstream MCP client, server-specific connection verification, platform compatibility details, and troubleshooting. The mutable latest tag and undocumented command-execution prerequisites add fragility.
Static review · not runListed 2026-07-29
Read the FMRS scoring method →Fit and risk
What it can accessReads local filesWrites / deletes local filesRuns commands or codeUses the network
Not for
- Environments where arbitrary command execution is prohibited
- Use cases requiring a documented tool set, authentication, or fine-grained permissions
Required permissions
- Permission to run arbitrary commands
- Optional access to the configured current working directory
- A local configured port for SSE and Streamable HTTP modes
Risks and side effects
- Arbitrary command execution can modify files, systems, or data
- An incorrect cwd can cause commands to run in an unintended directory
- Exposing an SSE or Streamable HTTP endpoint can broaden command invocation access
Setup
Before you start
Runtime：Node.js 22+
Use the npm package @agent-infra/mcp-server-commands. In stdio mode, configure cwd optionally. In SSE and Streamable HTTP modes, use npx and configure the port, which defaults to 8089.
Check that it works
The command-execution tools from @agent-infra/mcp-server-commands should appear in your client's tool list; asking the agent to run a simple command (like echo or a directory listing) and getting a result confirms the connection works.
Troubleshooting
- Verify the npm package identifier is @agent-infra/mcp-server-commands
- Check that cwd points to the intended working directory
- For SSE or Streamable HTTP, verify the configured port; the default is 8089
- Ensure the client uses the same transport configured for the server
Things to try
Once connected, you can ask your AI assistant things like:
- List all files in the current working directory
- Show me the installed Node.js version
- Run git status in this directory and summarize the result
- Create a file named test.txt containing hello
Use cases
Run commands in a configured working directory
Expose command execution through an MCP-compatible client
Overview
mcp-server-commands is an MCP server in the UI-TARS-desktop repository for running arbitrary commands. It is distributed as the npm package @agent-infra/mcp-server-commands and supports stdio, SSE, and Streamable HTTP transports. An optional current working directory can be configured; SSE and Streamable HTTP also require a port.
Similar servers
Context7
80 · B
Upstash's official server providing up-to-date third-party library docs for AI coding assistants
NPM Sentinel MCP Server
79 · B
AI-powered NPM package analysis MCP server
MartinLoop MCP Server
79 · B
An independent control layer that gives AI coding agents budgets, brakes, and receipts
GitHub MCP Server
76 · B
GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language
More servers from this repository
Source revision 3ac2cb8d946d Data synced 2026-10-19 Read the FMRS scoring method