Skip to main content
Glama
Sign In
codex-computer-use-mcp

codex-computer-use-mcp

by tmustier
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Local

js

Read-only

Executes JavaScript in OpenAI's cua_repl runtime to control native apps and browsers on a user's computer for Computer Use automation.

Instructions

Control native apps or browsers on the user's computer through OpenAI's official Computer Use JavaScript runtime (cua_repl). Computer Use is not ready: Codex app-server was not found. Checked: /Applications/ChatGPT.app/Contents/Resources/codex-cli/CodexCLI.app/Contents/MacOS/codex, /Applications/ChatGPT.app/Contents/Resources/codex, /root/Applications/ChatGPT.app/Contents/Resources/codex-cli/CodexCLI.app/Contents/MacOS/codex, /root/Applications/ChatGPT.app/Contents/Resources/codex. Install or update the official ChatGPT macOS app (https://chatgpt.com/download), then open it. If it is installed elsewhere, set "codexPath" to its absolute path in /app/config.json, or remove "codexPath" to restore automatic discovery.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "js"
  }
}

Input Schema

NameRequiredDescriptionDefault
codeNoJavaScript to run in the official cua_repl runtime.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Addedv0.7.1

TDQS

C2.2/5.0
Behavior1/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The description states the tool controls native apps and browsers by executing arbitrary JavaScript, which is inherently side-effecting, while annotations declare readOnlyHint=true and destructiveHint=false. The availability notice ("Computer Use is not ready", codexPath discovery) is genuinely useful context, but the safety-profile contradiction dominates.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness2/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The purpose sentence is front-loaded, but the body is a diagnostic dump enumerating four near-duplicate filesystem paths that an agent cannot act on, crowding out actual capability description. The remediation steps are useful once, but the path list should not be in a tool definition.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

With one param, no output schema, and annotations present, the bar is modest, yet the description leaves the agent unsure whether the tool is usable at all and what the runtime exposes. It also fails to explain the relationship to js_reset (resetting state) or computer_use_status (health check), which an agent would need to sequence calls correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Only one parameter exists and the schema already documents it at 100% coverage ("JavaScript to run in the official cua_repl runtime"). The description adds marginal meaning by naming the cua_repl runtime, but says nothing about what APIs are exposed, sandboxing, or return values of the executed code.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose3/5

Does the description clearly state what the tool does and how it differs from similar tools?

"Control native apps or browsers ... through OpenAI's official Computer Use JavaScript runtime (cua_repl)" gives a concrete verb and mechanism, and the 'code' param makes the JS-REPL nature clear. But the name 'js' is generic, none of the siblings (js_reset, computer_use_status) are referenced, and the majority of the text is an installation error rather than a statement of what the tool does.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

There is no when-to-use guidance: nothing says when to run a script here versus calling computer_use_status or js_reset. The only actionable guidance is remediation for the missing Codex app-server, which is an environment prerequisite rather than usage guidance.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools