How to Make Claude See Your Screen

You can let Claude see your screen by sharing a screenshot, by using Claude Desktop's screen context feature, or by enabling computer use in plans that support it. Tools like fenbs keep a written record of decisions and next steps beyond any visual session, so the context survives after the screen is no longer shared.

This guide explains each approach in detail, the permissions involved, and when each one is the right fit.

Sharing a screenshot: the quickest option

Take a screenshot and upload it to Claude.ai or Claude Desktop. Claude can read text in images, analyze layouts, describe what it sees, identify UI elements, and suggest actions based on the screen content. This is the simplest method and works in any plan that supports image uploads.

Use screenshots when you want Claude to look at a specific state: an error message, a UI design, a spreadsheet, or a form. The screenshot is static; Claude sees what was on screen at that moment and nothing more. This limitation is also its strength: you know exactly what Claude is working from.

Before sharing a screenshot, check that no passwords, private messages, or personal information are visible. Blur or cover sensitive areas if needed.

Screen context in Claude Desktop

On macOS, Claude Desktop can request screen context permission to read text and content from other apps without you taking a screenshot each time. This lets Claude see the active document, selected text in another app, or the name of the app you are using.

Grant this permission in Claude Desktop settings when prompted, and approve it in macOS System Settings under Privacy and Security, then Screen Recording or Accessibility. This does not let Claude control your apps; it can only read what is visible on screen.

Computer use: when Claude can act on your screen

Claude's computer use capability lets it view your screen and control a mouse and keyboard to take actions. This is available in specific Claude plans and requires explicit permission. Check Anthropic's current documentation for plan availability, as this changes.

Computer use is the most powerful of the three methods. Claude can navigate apps, fill forms, open files, and take multi-step actions. Before enabling it, decide clearly what it is allowed to do. A session with broad computer use and no written scope can produce changes that are difficult to reverse.

For comparison with ChatGPT's equivalent, see how to give ChatGPT access to your computer.

What permissions each method needs

• Screenshot upload: no special permission needed. Works in any Claude app that accepts image uploads.
• Screen context in Claude Desktop: grant Screen Recording or Accessibility permission in macOS System Settings when prompted.
• Computer use: enabled per session in Claude.ai on supported plans. May require Accessibility permission in macOS for input control.

When to use each method

Use a screenshot when you need Claude to look at one specific state and provide analysis or suggestions. It is the fastest method and requires no setup. Use screen context in Claude Desktop when you want Claude to reference what you have open without taking screenshots repeatedly during a working session.

Use computer use when you need Claude to take actions across multiple steps in an app and you have a clear scope defined. It is best suited to repeatable, well-understood tasks where each step can be reviewed. Avoid using it for tasks with irreversible consequences unless you have a rollback plan.

Screenshots are also the safest choice when the task involves sensitive information you want to review before sharing. Since you control what goes into the screenshot, you can crop out unrelated personal data or blur a field before uploading. Screen context and computer use share more broadly, so a screenshot gives you the most control over exactly what Claude sees.

What stays after the visual session ends

Screen access is session-bound. When you close the conversation, Claude no longer sees your screen and has no memory of what it observed unless you saved something. A design critique, a bug report, or a task list discussed visually is lost from context.

Writing the key decisions and next steps from a visual session into fenbs gives the next session something to work from. If Claude reviewed a design and flagged three changes, save those in a task so the next session starts with what was decided.

How to do this with fenbs

1. Create a free account. Your personal board is free and needs no credit card.
2. Add a project for the work that involves screen sessions. Write what the session is trying to accomplish.
3. After a session, record what was decided in Decisions and rules and what tasks were identified.
4. Connect Claude over MCP from Integrations. Ask Claude to read the project context at the start of the next session.
5. Before you finish each session, ask Claude to save a summary of what changed and what remains open.

fenbs does not capture or store screenshots. It keeps the structured record that makes visual sessions reusable rather than one-time events.