Skip to main content
City street used as an image CAPTCHA example

SCRAPINGBUDDY / CAPTCHA SOLVING

CAPTCHA Solving

Solve captcha challenges.
Keep automation moving.

Turn visual challenges into clicks, selections, and coordinates. Four focused solutions for your automation workflow.

IMAGE IN STRUCTURED ANSWER OUT

THE FIRST FOUR

A solution for each
kind of challenge.

Choose the image task that fits your workflow.
Get the answer in the format your application needs.

01 / 04

Every click, mapped.

Identify the points that match your instructions. Get x and y coordinates for each click.

YOU SEND
Challenge image + instructions
YOU RECEIVE
Click coordinates
Discuss your integration
CoordinatesILLUSTRATIVE EXAMPLE

Click coordinates / sample image

Sample city image with an illustrative answer overlay320, 200
EXAMPLE RESULT / 640 x 400 IMAGE
{ "coordinates": [{ "x": 320, "y": 200 }] }

FROM IMAGE TO ANSWER

Three steps.
One clear workflow.

Your application supplies the challenge.
The result tells it where to act.

01

Send the image

Choose a task type and include the image or pieces, along with clear instructions.

02

Resolve the challenge

The task is evaluated against your instructions to identify the matching points, tiles, or regions.

03

Use the answer

Receive the selection data. Your application applies the clicks, moves, or bounding boxes.

GOOD TO KNOW

A few answers
before you start.

Why do the returned coordinates click the wrong place?

Coordinates refer to the image you submitted, not its displayed size. Convert X by displayed width / submitted width, and Y by displayed height / submitted height. For example, X = 540 in a 1080-pixel image becomes X = 361 at 722 pixels wide. Add the image's left and top offsets for viewport clicks. These calculations assume the same uncropped image; account separately for cropping or padding and verify the challenge has not changed.

How do I keep grid tiles and drag-and-drop results aligned?

For grids, submit tiles left to right across each row, starting at the top; the tile count must equal rows multiplied by columns. Do not send both a combined image and a tile array. For Drag & Drop, preserve the input piece order because result entries correspond to those positions. Sorting pieces after submission or mixing coordinates from differently sized images can apply an otherwise valid answer incorrectly.

Can a slider puzzle use the Drag & Drop solution?

Compatible image-based puzzles can use a background and a separate slider-piece image. The returned destination is not a ready-to-use drag distance: your application must account for the piece's starting position and displayed scale. There is no dedicated slider solution here, and finding the target position does not guarantee the website will accept the interaction.

Why is my file rejected when the image itself is within the size limit?

File limits apply to decoded bytes, but the complete JSON request has a separate 5,000,000-byte limit. Base64 adds roughly one third to the original size, and combined tiles or pieces can push the request over that limit. Send file bytes encoded as Base64 or a matching data URL, not a filename or remote image URL. Check the chosen task's file and dimension limits before submitting.

My request timed out while the task was processing. Should I resubmit?

Continue checking the saved task instead of creating a replacement. A client timeout does not cancel server-side work. Preserve the original Idempotency-Key and input: an identical replay returns the saved task, while a different key can create another paid task. If submission is unconfirmed, contact support with the task ID. Poll at five-to-ten-second intervals and coordinate polling across workers.

Why is credit still reserved, or a ready answer waiting for credit?

A reservation is not the final charge. Confirmed failed tasks release it; a successful task settles against actual provider cost. A timeout can leave the outcome unresolved, so the hold remains pending review. If billingStatus is review, contact support with the ID. If it is awaiting_credit, add credit to receive the saved answer rather than paying to create another task.

The answer looks correct, but the website rejected it. What should I check?

Check that you applied the answer to the same, still-current challenge, with the correct image scale, tile order or piece mapping. These image solutions return answer data, not a website acceptance decision or universal verification token. Preserve the task ID and a redacted description of the mismatch for support; do not assume a ready task proves the website accepted the result.

BUILD YOUR NEXT WORKFLOW

Your next step,
without the puzzle.

Tell us what you need to solve. We will help you find the right fit.

Talk to our team