video-link-extract

Extract Copy from Short Video Links

📝 Introduction

Extract Copy from Short Video Links lets you automatically get every sentence spoken in a video with just one video link. Whether it's the lines from a trending Douyin video, the explanation content of a Xiaohongshu tutorial, or the spoken copy of a Kuaishou livestream, paste a link and you can get the complete text version, along with the timestamp where each sentence appears.

Suitable for bloggers doing content re-creation to break down viral scripts, operations staff analyzing competitor video scripts, and video editors quickly extracting subtitle materials.

✨ Features

Core Features

Feature Module	Capability Description	Core Value
Multi-platform Support	Automatically recognizes video share links from platforms such as Douyin, Kuaishou, and Xiaohongshu	One tool covers mainstream short video platforms, no need to switch back and forth
Intelligent Recognition	AI accurately recognizes speech content in videos and converts it to text	High recognition accuracy, stable output even with dialects and noisy environments
Sentence Segmentation	Each sentence is marked with start and end times, accurate to the millisecond	Quickly locate where a sentence appears in the video, convenient for citation and editing
Asynchronous Waiting	Automatically tracks processing progress after submission, no manual refresh needed	Automatic background processing, returns results immediately upon completion
Error Prompts	Provides clear prompts when encountering issues such as unsupported platforms or overly long videos	Quickly locate the cause of the problem, no need to guess why it failed

Highlights

• Zero-threshold Operation: Paste the link to get the copy, no need to download the video or install transcription software
• Millisecond-level Timeline: Each sentence is accurately marked with start and end times, directly usable for subtitle creation
• Automatic Progress Tracking: No need to wait around after submission, intelligent polling automatically obtains the final result
• Complete Error Explanations: Issues such as invalid links or unsupported platforms are clear at a glance, with suggested solutions

📖 Usage Guide

🔑 Authentication

Go to Red Fox Hub to get the API KEY

Configuration method:

1. Configure in ~/.openclaw/openclaw.json Configure in
2. Set environment variable REDFOX_API_KEY
3. Use command-line parameter --api-key

📱 How to Get Video Links on Each Platform

Before using this skill, you need to copy the video link from each platform's App. The following are the specific operation methods for the three major platforms:

Douyin

How to get the link:

1. Open the Douyin App and find the video whose copy you want to extract
2. Tap the "Share" button in the lower right corner (arrow icon)
3. In the share panel, select "Copy Link"

Supported link formats:

• Share short link:https://v.douyin.com/xxxxxx/
• Web link:https://www.douyin.com/video/xxxxxx

Platform limitations:

• ⚠️ Only supportspublic videos. Private or deleted videos cannot be extracted
• ⚠️ Image-text posts (without a video stream) are not supported
• ⚠️ Video duration cannot exceed 10 minutes

Kuaishou

How to get the link:

1. Open the Kuaishou App and find the video from which you want to extract the text
2. Tap the "Share" button at the bottom right
3. In the share panel, select "Copy Link"

Supported link formats:

• Share short link:https://v.kuaishou.com/xxxxxx
• Web link:https://www.kuaishou.com/short-video/xxxxxx

Platform limitations:

• ⚠️ Only supportspublic videos. Private or deleted videos cannot be extracted
• ⚠️ Image-text posts (without a video stream) are not supported
• ⚠️ Video duration cannot exceed 10 minutes

Xiaohongshu

How to get the link:

1. Open the Xiaohongshu App and find the video note from which you want to extract the text
2. Tap the "Share" button (arrow icon) at the top right
3. In the share panel, select "Copy Link"

Supported link formats:

• Web link:https://www.xiaohongshu.com/discovery/item/xxxxxx
• Share short link: the short link copied from within the Xiaohongshu App

Platform limitations:

• ⚠️ Only supportsvideo notes. Image-text notes (pure images/text) have no video to extract
• ⚠️ The note must be inpublic status. Private notes cannot be accessed
• ⚠️ Video duration cannot exceed 10 minutes

💡 Tip: If you're not sure whether the copied link will work, just paste it to me and I'll check it for you.

Quick reference for common phrases

Intent	Example phrase	Effect
Extract video text	"Help me extract the text from this Douyin video"	Enter a link and automatically return the complete text version
Get subtitles with a timeline	"Extract the text from this video, with timestamps"	Output details for each sentence (minutesin minutes:seconds format for start and end times, all listed), can be used directly as subtitles
Analyze video script	"Analyze what this viral video says"	Get the complete text to help break down the script structure

🔄 Workflow

1. Enter video link: Provide a video share link from platforms such as Douyin, Kuaishou, or Xiaohongshu
2. Submit processing task: The system automatically submits the text extraction task and returns a task identifier
3. Smart wait for processing: Automatically tracks task progress, checking every few seconds until it is complete
4. Return extraction results: Output the complete segmented text + video content summary, with details for each sentence generated as a separate .md file for click-to-view

For the complete execution process, see references/core_workflow.md

🔗 API Documentation

API Endpoint

API Name	Request Method	API Endpoint
Submit Transcript Extraction Task	POST	https://redfox.hk/story/api/parseWork/audioTextExtract/submit
Query Extraction Result	POST	https://redfox.hk/story/api/parseWork/audioTextExtract/result

Submit Task - Request Parameters

Parameter Name	Type	Required	Description
url	string	Yes	Video link (supports Douyin/Kuaishou/Xiaohongshu share links or web links)
source	string	Yes	API source identifier, fixed value:"短视频链接提文案-GitHub"

Submit Task - Response Structure

Field Name	Type	Description
taskId	string	Task identifier, used for subsequent querying of the extraction result

Query Result - Request Parameters

Parameter Name	Type	Required	Description
taskId	string	Yes	Task identifier returned when submitting the task
source	string	Yes	API source identifier, fixed value:"短视频链接提文案-GitHub"

Query Result - Response Structure

Field Name	Type	Description
taskId	string	Task identifier
status	string	Task status: queued / running / succeeded / failed
text	string	Full extracted transcript content (only has a value on success)
stampSents	list	List of details for each sentence (only has a value on success), each item contains textSeg (text segment), start (start time in ms), end (end time in ms)
failReason	string	Failure reason (only has a value on failure)

Data Scope

Data Type	Supported Range	Description
Platform Support	Douyin, Kuaishou, Xiaohongshu	A video share link or web link from the corresponding platform is required
Video Duration	No more than 10 minutes	For overly long videos, please edit first and then extract
Content Type	Only video works are supported	Image-text works cannot be extracted
Video Visibility	Public videos	Private or deleted videos cannot be extracted
Return Precision	Millisecond-level timestamps	Each transcript sentence is annotated with precise start and end times

Platform-Specific Notes

Platform	Where to Get the Link	Common Issues
Douyin	In the App, on the video page → Share button → Copy Link	Videos posted by private accounts cannot be extracted; some livestream replays may not be supported
Kuaishou	In the App, on the video page → Share button → Copy Link	Same as Douyin, only public videos are supported
Xiaohongshu	In the App, on the note page → Share in the upper right corner → Copy Link	Please confirm that the note is a video type (the cover has a play icon); image-text notes are not supported

Fallback Handling

When the target data is not found or processing fails:

1. Link parsing failed → Prompt the user to check whether the link is valid and whether it comes from a supported platform
2. Video timeout → Prompt the user that the video must be within 10 minutes and suggest cropping it first
3. Image-text type → Prompt that the link is an image-text post and there is no video to extract
4. System busy → Prompt the user to try again later; the system will recover automatically

⛔ Important: When the API returns an exception or extraction fails, directly show the user the API's original error message,must notaccess the link content through a browser or other means as a fallback.

💰 Credit refund: When extraction fails, also inform the user that the deducted credits have been returned to the current account and they can go to Credit Records to view the details.

❓ FAQ

Usage-related

Q: Which platforms' videos are supported? A: Currently supports video links from three major platforms: Douyin, Kuaishou, and Xiaohongshu. Simply paste the share link or webpage link.

Q: What is the maximum video length? A: Supports videos up to 10 minutes long. If the video exceeds 10 minutes, it is recommended to crop it with an editing tool before extracting.

Q: What does the extraction result include? A: It includes three parts — the complete transcript segmented by semantics and a brief summary of the video content displayed directly in the conversation, and each sentence's details (inminute-second format start and end times, all listed) generated as a separate .md file; click to view the complete content.

Q: Can image-text content be extracted? A: No. This skill is specifically for extracting speech content from videos. Image-text posts have no video stream and cannot be processed.

Troubleshooting

Q: What should I do if it says "link parsing failed"? A: Please confirm that the link is from Douyin, Kuaishou, or Xiaohongshu and is a video type (not image-text). Both share links and webpage links are supported.

Q: Waiting for a long time with no result returned? A: Video processing takes some time, and the system will automatically wait and track progress. If there is still no result after more than 2 minutes, you can resubmit and try again.

Q: What should I do if the extracted transcript has typos? A: Speech recognition may produce a small number of deviations in noisy environments or dialect scenarios. It is recommended to manually proofread key content. The overall accuracy rate is above 95%.

📚 Reference Documentation

• Detailed execution workflow:references/core_workflow.md