OpenAI Codex Gets Appshots and Remote Mac Control: What Indie Hackers Need to Know

OpenAI has established a weekly shipping cadence called Codex Thursday. Every Thursday they push updates to Codex. Yesterday's drop was the most significant one yet.

Three features shipped: Appshots for macOS, remote locked Mac access from iPhone, and Goal mode reaching general availability. Here is what each one actually does and whether it changes your workflow.

What Just Shipped on Codex Thursday

flowchart TD
    A[iPhone: start task] --> B[Appshots: Command+Command captures context]
    B --> C[Goal mode: runs autonomously for hours]
    C --> D[Mac continues even when locked]
    D --> E[iPhone: monitor progress remotely]
    E --> F[Review and approve from anywhere]

Appshots: Context Without Copy-Pasting

The friction point that every developer hits with AI coding tools is context. You are looking at an error in your browser. You switch to Codex. Now you have to describe the error, copy the stack trace, paste the URL, explain what you were doing. Every switch between tools bleeds context.

Appshots eliminates that friction. Press both Command keys to send the frontmost app window to Codex with a screenshot and available text, so Codex can work from context in another app without you copying, pasting, or describing it manually.

The keyboard shortcut is Command+Command (both Command keys simultaneously). Codex receives a full screenshot of the frontmost window plus any available text content. If you are looking at a browser error, a terminal output, a design file, or a GitHub PR, one keypress gives Codex everything it needs to understand the task.

You can also configure a custom hotkey in Codex preferences if Command+Command conflicts with something in your setup.

What this changes in practice: The round-trip between your working context and Codex drops from 30-60 seconds of manual explanation to a single keystroke. For developers who context-switch dozens of times per day between their editor, browser, and terminal, this compounds into real time savings.

The limitation: Appshots arrived as part of Codex app build 26.519 and is currently Mac-only. Windows and Linux developers cannot use this yet.

Remote Mac Control: Codex While Your Screen is Off

This one is genuinely new territory. Codex can now continue using desktop apps even after a Mac has been locked. You trigger and monitor this from Codex Mobile on your iPhone.

The practical scenario: you start a long Codex task before leaving your desk. Your Mac screen locks after a few minutes of inactivity. Previously, Codex would pause or lose context. Now it keeps working. You check in from your phone, review what it has done, approve the next steps, and keep moving.

OpenAI explains that Codex scopes locked use to active, trusted computer use turns and includes safeguards such as short-lived authorization, covered displays, relock on local input, and manual-unlock fallback.

The security model is careful. Short-lived authorization means the locked access window expires automatically. Covered displays means the screen stays dark while Codex works. Relock triggers when you physically interact with the machine. These are not afterthoughts.

Setup involves updating both the ChatGPT mobile app and the Codex Mac app, then pairing the two by scanning a QR code. Once paired, you get full Codex monitoring from your phone.

What this changes in practice: Long autonomous tasks no longer require your Mac to stay awake and unlocked. For indie hackers running overnight test suites, large refactors, or multi-step build processes, this removes a real constraint.

The limitation: Remote locked access is Mac-only. Support for connecting your phone to Codex on Windows is coming soon but is not available today.

Goal Mode is Now GA

Goal mode is no longer an experimental feature and is available in the Codex app, IDE extension, and CLI. With Goal mode, you can have Codex drive toward a specific objective for hours or even days.

The previous /goal command was experimental and limited. GA means it is production-ready and available everywhere: the desktop app, the IDE extensions (VS Code, JetBrains), and the CLI. The behavior is unchanged: you set a goal, Codex works until it reaches it. What changes is the reliability and the availability across all entry points.

You can check in and steer, and even pause Codex along the way. Pro tip: start side chats to understand the work that has been done so far without having to interrupt the main task.

Side chats are underrated here. You can ask "what have you done so far?" or "why did you make this decision?" in a separate thread while Goal mode keeps running in the main thread. No interruption, full visibility.

What This Means for Codex vs Claude Code

The honest assessment: these features are meaningful but Mac-specific.

If you develop on Mac, Appshots and remote locked access are genuinely useful additions that Claude Code does not have equivalents for today. The context capture workflow with Appshots is faster than anything Claude Code currently ships. The locked Mac access enables autonomous task patterns that were not previously reliable.

If you develop on Windows or Linux, nothing in this Codex Thursday update applies to you yet. Claude Code is cross-platform. Codex's best new features are not. That is a real competitive gap that matters for non-Mac developers until OpenAI ships Windows parity.

Goal mode GA affects all platforms equally and is the update most relevant to developers regardless of OS. The Codex /goal command comparison with Claude Code agents covers the autonomous task approach in detail if you want the deeper breakdown.

For the broader question of whether Codex or Claude Code is the right fit for your stack, the Codex vs Claude Code comparison and the Codex mobile launch breakdown from last week are worth reading alongside this update.

OpenAI is shipping fast. Codex Thursday gives them a predictable cadence to announce features, build community anticipation, and compete week-by-week with Anthropic's Claude Code releases. For indie hackers tracking which AI coding tool to invest in, paying attention to that cadence is worth the few minutes it takes each Thursday.