Free tools Windows power users keep installed
One-click scans. No signup required.
To share a screenshot with an AI agent, attach the image in the chat composer, drag it into the conversation, or paste it from your clipboard. Then tell the agent what the image shows, which area matters, and what you want done. If the agent must operate a live desktop or browser, use a computer-use integration that captures screenshots during an action loop rather than sending one static image.
The three ways to give an AI agent visual context
- Upload a still image. Use the plus or attachment control, drag the file into the prompt, or paste a copied screenshot. This is best for explaining an error, reviewing a design, extracting visible text, or comparing two states.
- Pass an image from a command line. Codex supports image arguments such as
codex -i screenshot.png "Explain this error and suggest the smallest fix". For a before-and-after review, usecodex --image before.png,after.png "Compare these states and list the regressions". Command-line syntax and availability can vary by installed CLI version, so check the version on your machine if a flag is rejected. - Enable computer use. A computer-use tool lets the agent request screenshots, clicks, typing, or other actions. Your application executes those requests and returns the resulting screen image. This is an interaction loop, not the same permission model as uploading one picture.
Upload a screenshot in ChatGPT
In the web composer, open the plus menu and choose photos or files. You can also drag an image into the prompt area or paste one copied to the clipboard. The current ChatGPT image-input help page lists PNG, JPEG, and non-animated GIF files, with a 20 MB limit per image. It does not promise one fixed number of images for every conversation; practical capacity depends on image size and the text you send with it.
After attaching the file, include a focused instruction. For example: “This is the checkout page after clicking Pay. Inspect the red message beside the card field, transcribe it exactly, and suggest the smallest code change that could cause it.” If text is tiny, crop or enlarge the relevant area before uploading. An unclear image can produce an uncertain interpretation, and image performance can vary for specialized medical imagery or non-Latin text.
Upload a screenshot in Claude
Claude provides an Add files or photos control, plus drag-and-drop and clipboard paste. Its help documentation lists JPEG, PNG, GIF, and WebP. It documents up to 20 files per chat, a 500 MB limit per uploaded file, and image dimensions up to 8000 by 8000 pixels for chat uploads. These are platform documentation limits and may change.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errors#1 Best Overall
Claude’s PDF handling is different from uploading an individual image: PDFs up to 100 pages can be analyzed for text and visual elements, while pages 101–1000 are processed as text only. If a screenshot is embedded on page 150 of a long PDF, upload the image itself when visual inspection is required.
Use Codex with an image file
For a terminal workflow, place the screenshot in your working directory and pass it with the image option:
codex -i screenshot.png "Explain this error and suggest the smallest fix"To compare two states, provide both files and define the comparison:
codex --image before.png,after.png "Compare these states and list the regressions"Use common formats such as PNG or JPEG. If your installed Codex command reports an unknown option, run its local help or version command; command-line interfaces can change independently of the documentation example.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Write a prompt that produces a useful inspection
A screenshot rarely contains enough context by itself. Put four pieces in the same message:
- Identity: name the application, page, dialog, or state shown.
- Focus: point to the panel, control, warning, error, or visual difference to inspect.
- Task: say whether you want an explanation, exact transcription, accessibility review, comparison, diagnosis, or a proposed fix.
- Boundaries: state what not to infer, such as “Do not guess text hidden outside the crop; list unreadable words as uncertain.”
A reusable one-image prompt is:
This screenshot shows [application/page and state]. Inspect [specific region]. Please [desired output]. Quote visible text exactly, separate observations from inferences, and mark anything unreadable instead of guessing.For two or more images, identify each by filename or order and define the relationship:
Image 1 is before saving; image 2 is after. Compare only visible regressions. Check the warning banner first, then list any changed controls or text.If you need a sequence extracted from several captures, say “Describe the steps in order” rather than asking for a vague summary.
Make the screenshot legible without losing context
- Capture sharply: avoid motion blur and compression artifacts. If text matters, make it large enough to read at normal zoom.
- Crop deliberately: remove unrelated windows and empty margins, but leave enough surrounding labels to identify what a control belongs to.
- Resize with care: very large images may be resized by the service before processing, which can make small text harder to read. Cropping the relevant section often preserves more useful detail than uploading a giant full desktop.
- Keep state visible: include the page title, selected tab, or nearby status indicator when it explains the error.
- Do not depend on coordinates: pixel positions are not portable across display dimensions, scaling, or computer-use environments.
Still screenshots versus live computer use
A manual upload asks one visual question about a captured state. Computer use supplies a tool that the agent can call repeatedly. In the documented Claude computer-use cycle, you provide the computer tool and a user request; the model returns tool calls such as screenshot, click, or type; your application executes them; you return the result, including screenshot images for screenshot actions; and the model continues until the task is complete.
The model does not directly connect to your laptop or server. Your integration controls which applications are exposed and performs the requested actions. OpenAI’s computer-use guidance similarly describes an approved-application workflow with permission prompts and scoped access. Use a still image for a one-off review; use computer use only when the agent must observe changing state or perform approved actions.
Privacy and safety before you upload
A screenshot shares everything visible in the captured frame, not only the area named in your prompt. Inspect the entire image and crop or redact information that is unnecessary:
Rank #3
- passwords, recovery codes, API keys, and session tokens;
- personal messages, customer records, health information, or financial details;
- unrelated browser tabs, notifications, and internal URLs.
Review the current data-use and retention controls for the account and product you use. OpenAI documents product-specific choices and states that ChatGPT Enterprise content is not used to train models. Do not generalize that statement to every ChatGPT plan.
Live interaction adds risk. Anthropic warns that screenshots, webpages, and application interfaces can contain deceptive instructions or prompt-injection text. Limit the agent to the applications and actions it needs, require a person to confirm consequential actions such as purchases or sending messages, and keep an action log. Treat files opened inside the target application as data the agent may process, not as trusted instructions.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchCommon problems and fixes
The attachment control is missing
Check that you are using a plan, workspace, or client that supports image input, then reload the page or update the app. If the control remains unavailable, use a supported web client or the product’s documented API/CLI path instead of assuming the feature is enabled for every account.
The upload is rejected for size or format
Convert to PNG or JPEG for broad compatibility and reduce dimensions or file size while keeping text readable. ChatGPT’s documented per-image limit is 20 MB. Claude lists JPEG, PNG, GIF, and WebP, with up to 500 MB per uploaded file and dimensions up to 8000 by 8000 pixels.
The agent misreads small text
Crop to the relevant panel, enlarge it, and upload the crop alongside a contextual view if necessary. Ask the agent to mark uncertain characters rather than infer them. A second, tighter crop is usually more useful than repeating the same full-screen image.
The agent answers the wrong question
Rewrite the prompt with an explicit focus and output format. “Review this” is underspecified; “List only regressions between Image 1 and Image 2, quoting changed labels exactly” gives the model a testable task.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →A PDF screenshot is not analyzed visually
Extract and upload the page image directly. In Claude’s documented behavior, pages 101–1000 of a long PDF are processed as text only, so an embedded screenshot on those pages may not receive visual analysis.
Computer-use actions are unsafe or unexpected
Reduce the exposed application set, remove write or purchase permissions where possible, add confirmation gates for external side effects, and stop the loop when the visual state differs from the expected one. Never allow a page’s on-screen instruction to override your own task policy.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Automate website screenshots for an agent
If the image you need is a public webpage rather than your local desktop, an API can create the capture before your agent receives it. ScreenshotNeo is a website screenshot API and MCP server. It is useful when you want a repeatable URL-to-image step instead of browser setup: it accepts consent banners like a visitor, removes more than 60 known consent platforms plus newsletter popups and chat widgets, and lets you turn each cleanup step off.
Only clean shots are billed. Bot checks or CAPTCHAs, blank pages, timeouts, failed loads, and cache hits cost nothing, and each response reports the outcome with X-Page-Verdict and X-Billed headers. Its MCP server exposes take_screenshot, get_page_info, and capture_pdf for Claude, Cursor, and other MCP clients.
Recommended Free Tools
One-call capture with cURL
See the parameter details in the ScreenshotNeo documentation:
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webpPython
import requests
r = requests.get("https://api.screenshotneo.com/v1/shot", params={"access_key": "YOUR_API_KEY", "url": "https://stripe.com"}, timeout=90)
open("shot.webp", "wb").write(r.content)Node.js
const q = new URLSearchParams({ access_key: 'YOUR_API_KEY', url: 'https://stripe.com' });
const res = await fetch(`https://api.screenshotneo.com/v1/shot?${q}`);The service has 63 options for production workflows: full-page capture with lazy images loaded; CSS-selector element capture; dark mode; 12 device presets or any viewport; retina scale; PDF paper size, margins, landscape, and page ranges; HTML/CSS-to-image rendering; custom CSS and JavaScript; pre-capture clicks; hidden selectors; waits for a selector, delay, or network idle; blocking ads, trackers, requests, or resource types; custom headers, cookies, user agents, and Authorization; timezone and geolocation; transparent backgrounds; image resizing; configurable-TTL caching; signed links for public image tags; asynchronous jobs with signed webhooks; bulk capture of up to 100 URLs per call; a usage API; an OpenAPI specification; and compatibility with parameter names used by other screenshot APIs.
Plans are Free with 1,000 shots per month and no card, Starter at $5 for 3,000, Growth at $15 for 15,000, Pro at $39 for 60,000, Scale at $99 for 250,000, and Business at $249 for 1,000,000. Yearly billing gives two months free, and every feature is included on every plan.
Or skip the browser setup
Use the call above when an agent needs a clean webpage image: cookie banners, newsletter popups, and chat widgets are removed before the shot; bot checks, blank pages, and failed loads are never billed; and the MCP server lets AI agents request screenshots directly. You can start with 1,000 screenshots a month free, with no card, and paid plans start at $5 for 3,000. Create a free ScreenshotNeo account.
A repeatable screenshot-to-agent workflow
- Capture only the application or page state relevant to the task.
- Inspect the full frame and redact secrets or unrelated personal data.
- Crop or resize until important text is legible while preserving identifying context.
- Attach the image, or generate a webpage capture with an API.
- State the image identity, focus area, requested action, and uncertainty rule.
- For multiple images, label each file and define the exact comparison or sequence.
- Check the response against the visible evidence before applying a suggested fix.
- For live computer use, keep permissions narrow and approve consequential actions manually.
Frequently Asked Questions
Can an AI agent read a screenshot that contains handwriting?
It may attempt to interpret it, but accuracy depends on contrast, resolution, and legibility. Capture a close, well-lit image and ask the agent to mark uncertain words rather than silently correcting them.
Should I send a full desktop or only the application window?
Send the smallest frame that still explains the task. Include nearby titles, tabs, or status indicators when they provide necessary context; otherwise crop them out to reduce accidental disclosure.
Can I use one screenshot for an automated regression test?
A single image documents one state. Reliable regression checking requires consistent captures before and after the change, identical viewport conditions, and a prompt or program that defines which visual differences count.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
