HelloFont Chinese Font Recognition

Problem

When an image contains only Chinese text but no font metadata, identifying the typeface usually relies on visual guessing or manual comparison against a font list. This skill turns the question of which font is used in an image into a structured recognition workflow: OCR the text, return the best match and candidate fonts, and render a locally openable HTML result.

How It Works

The input must be a JPEG, PNG, or WebP image. The skill expects the image to be clear and tightly cropped to the text region; if an attachment is already present, it should be used directly instead of requesting re-upload. Each recognition intent reuses one client_request_id to avoid concurrent duplicate requests. After calling the recognition endpoint, the flow may receive HTTP 402, enter a payment confirmation step, and then replay the result using the original out_trade_no after successful payment, rather than creating another order. A successful result requires HTTP 200 with mode=real. The local renderer then produces hellofont-result.html, showing the original image, OCR text lines, the best match best_font, candidate fonts, and confidence. If WebFont loading fails, the page displays a system-font fallback.

Boundaries and Notes

The skill only outputs font matches and OCR text; it does not determine font copyright, licensing, or commercial-use eligibility. Font matches and confidence are reference values, not accuracy guarantees. NO_TEXT means no text was detected and is not an interface failure. The current API does not provide verifiable text coordinates, so the result page does not draw bounding boxes. If payment is not successful, payment status is unknown, or the response does not return mode=real, the recognition should not be claimed as completed.