ChatGPT Web Now Supports Skills – Unlocking a Hidden Cloud Computer

The author notes that the latest ChatGPT web interface has merged Codex‑like Skills and a cloud‑based sandbox, allowing users to upload and run over a hundred Skills without consuming Codex limits, effectively turning the service into an always‑on AI “cloud computer” for tasks such as file handling, document generation, and web automation.

Old Zhang's AI Learning
Old Zhang's AI Learning
Old Zhang's AI Learning
ChatGPT Web Now Supports Skills – Unlocking a Hidden Cloud Computer

Recent updates have collapsed the three original modes (Chat, Work, Codex) into just ChatGPT and Codex, with the ChatGPT mode further divided into Chat and Work.

The web interface now looks very similar to the Codex client, as shown in the side‑by‑side screenshots.

Earlier posts covered the file library, projects, and scheduled tasks that existed before the Work mode and when plugins were still called Apps. Today the plugin system and Skills appear almost identical to Codex.

Skills can be created in three ways; the author prefers uploading Skills directly from the desktop client. The second method, the Skills editor, also supports file attachments and improved descriptions.

Plan: migrate the existing 100+ desktop Skills to the web, moving a subset to run online. Benefits include (1) no consumption of Codex quota and (2) resilience to the local computer being shut down, especially when combined with web‑based scheduled tasks.

Usage: switch the ChatGPT web UI to Work mode and invoke a Skill with @skills.

The Work mode runs in a cloud sandbox that provides several VM‑like capabilities:

Python execution : a stateful Jupyter Notebook environment for running Python, handling files, and generating charts.

Web operation : a remote Cloud Browser that can access and manipulate web pages.

HTML/React preview : sandboxed rendering of front‑end code.

Work cloud tasks : cloud‑hosted analysis, document, spreadsheet, PPT, and website generation, backed by an isolated compute environment.

For a Plus user, the sandbox provides a 4‑core CPU, 4 GB RAM, about 60 GB temporary disk, and a Linux container without GPU.

The author uploaded a frequently used "z‑web‑pack" Skill to the web. The Skill runs in the cloud computer at the path /root/.codex/skills/remote-skills, and the resulting output is a compressed package that can be downloaded locally.

Initially the author thought OpenAI merely transplanted part of Codex into the ChatGPT web UI, but the growing set of features—file library, projects, scheduled tasks, plugins, Skills, cloud sandbox, web operations, and document generation—are converging onto a single workbench. The boundary between the web UI and the Codex client is becoming increasingly blurred.

Codex remains better suited for heavy programming and long‑running local projects, yet for most users the ChatGPT web version is evolving into an easier‑to‑use "cloud Codex": no environment setup, no dependence on a powered‑on PC, and the ability to run uploaded data and Skills entirely in the cloud.

This trend aligns with OpenAI’s recent acquisition of a company that enables agents to run continuously in the background, suggesting a strategy to merge chat, office, coding, and automation into a single entry point.

In the future, users may open ChatGPT not just to ask a question but to hand over an entire workflow—letting the system produce files, webpages, tables, code, and even full websites.

Original Source

Signed-in readers can open the original source through BestHub's protected redirect.

Sign in to view source
Republication Notice

This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactadmin@besthub.devand we will review it promptly.

ChatGPTOpenAIAI AutomationSkillsCloud SandboxCodex
Old Zhang's AI Learning
Written by

Old Zhang's AI Learning

AI practitioner specializing in large-model evaluation and on-premise deployment, agents, AI programming, Vibe Coding, general AI, and broader tech trends, with daily original technical articles.

0 followers
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.