skills.rest
shalevamin
shalevaminCOMMUNITY
Explore all skills by this author →
GitHub

tua-openai-operator

Runs Tau's local OpenAI operator console for GPT-5.4 computer-use browser automation.

Updated Mar 7, 2026
Search Tags:#browser automation#playwright#openai#computer-use#replay#gpt-5.4#operator console
One-click install
npx skills add https://github.com/shalevamin/Tau_agent --skill tua-openai-operator-shalevamin
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: tua-openai-operator
Source: https://github.com/shalevamin/Tau_agent/tree/main/tau-agent-main/skills/tua-openai-operator
Command: npx skills add https://github.com/shalevamin/Tau_agent --skill tua-openai-operator-shalevamin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Ad-hoc browser automation loops lack visibility, replay, and step-by-step validation. This Skill launches Tau's bundled OpenAI CUA sample app so browser tasks run through a dedicated operator console with screenshots, event streaming, and replay artifacts. ## Core Features & Use Cases - Dedicated Operator Console: Start the local OpenAI CUA stack via tua-cua-dev, tua-cua-runner, or tua-cua-web launchers. - Dual Execution Modes: Use native mode for direct computer-use actions or code mode for deterministic Playwright scripting in a browser REPL. - Verifiable Runs: Review screenshots, activity logs, and replay artifacts before marking a browser task complete. - Use Case: A user needs a multi-step web workflow validated visually; the operator console runs it with GPT-5.4 computer-use and produces replayable evidence of each step. ## Quick Start Start the Tau OpenAI operator console and run this browser task in native computer-use mode with screenshots and replay enabled.

Frequently Asked Questions about tua-openai-operator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run OpenAI computer-use browser automation locally?▼

Start the bundled OpenAI CUA sample app using the tua-cua-dev, tua-cua-runner, or tua-cua-web launcher commands. The stack runs locally at ~/.openclaw/tools/openai-cua-sample-app and requires an OPENAI_API_KEY.

What is the difference between native and code mode in computer-use?▼

Native mode sends direct computer-use actions like clicks and keystrokes through the OpenAI Responses API. Code mode uses a persistent JavaScript REPL over Playwright, which is more deterministic for scripted browser interactions.

When should I use the OpenAI operator console instead of browser-use?▼

Use the operator console when a task needs screenshots, replay, event streaming, or step-by-step validation. Tau still prefers browser-use for ordinary browser work, escalating to this console only for controlled, verifiable runs.

Which model does the OpenAI computer-use stack require?▼

The stack uses OpenAI-family models only, preferring gpt-5.4 for native computer-use runs. A single OPENAI_API_KEY environment variable is the only required credential.

Can I use an already-open personal browser session with this console?▼

No, the operator console works against a controlled web workspace. For tasks needing an existing personal browser session, escalate to the tua-computer-use skill before using extension relay.

Related Skills

Explore More Software Engineering Skills→
QwenLMQwenLM
27.5k

cua-driver

Automate desktop and browser applications in the background without stealing focus.

Official
Advanced
MadAppGangMadAppGang
7

proof-of-work

Generate screenshots, test results, and deployment logs with confidence scoring.

Official
Intermediate
T-0-coT-0-co

autonomous-longtask

placeholder

Official
Advanced
westnwestn
1

open-gstack-browser

Launch a headed Chromium browser with the GStack sidebar extension loaded.

Community
Advanced