bestblogs-process-videos

BestBlogs Video Batch Transcription and Content Update Workflow

Skill Overview

The BestBlogs Video Batch Transcription and Content Update Workflow is an automated tool that helps BestBlogs administrators batch-query and transcribe YouTube videos, then update them in the content library. It supports automatically selecting transcription quality based on priority and provides a streamlined batch-processing mode.

Applicable Scenarios

Batch Content Updates: Use serial processing when a large number of YouTube videos need to be transcribed and added to the BestBlogs content library, ensuring that each video ID is correctly associated with its content.

Intelligent Quality Selection: Automatically select a transcription mode based on video priority. High-priority videos use pro mode for higher-quality transcription, while others use think mode for faster processing.

Streamlined Batch Processing: When processing a large number of videos, use streamlined mode to process all videos or the first N videos directly without individual confirmation, improving operational efficiency.

Core Features

Four-Stage Automated Workflow: Query the list of videos awaiting preprocessing → user selection (which can be skipped in streamlined mode) → process videos one by one (transcribe, save, update, and trigger analysis) → output results. Each stage runs independently, with clearly visible status information.

Intelligent Transcription Quality Control: Automatically map transcription modes according to video priority. HIGH-priority videos use pro mode to call Gemini Pro for higher-quality transcription, while MEDIUM/LOW/NONE-priority videos use think mode for faster processing. If pro mode times out, the system automatically falls back to think mode and retries.

Data Security: Videos are processed strictly in sequence, locking the association between each video’s ID and URL. File names include the ID as a reference credential. Parallel processing and cross-ID updates are prohibited, ensuring that each transcription corresponds exactly to its video record.

Frequently Asked Questions

How do I configure authentication for BestBlogs video transcription?

You need to set two environment variables: BESTBLOGS_ADMIN_USER_ID (the administrator user ID) and BESTBLOGS_ADMIN_JWT_TOKEN (the administrator JWT token). These variables are sent as HTTP request headers to the https://api.bestblogs.dev API endpoint. If you encounter a 401 or 403 error, the token may have expired or you may not have sufficient permissions, in which case you need to update the authentication information.

What is the difference between pro mode and think mode for video transcription?

Pro mode uses Gemini Pro for in-depth reasoning. It provides higher transcription quality but takes longer, making it suitable for high-priority (HIGH) video content. Think mode uses the standard transcription mode and processes videos more quickly, making it suitable for medium- and low-priority videos. If pro mode fails due to a Chrome AppleScript timeout (ETIMEDOUT), the system automatically falls back to think mode and retries once.

How is data security ensured when processing videos in batches?

The workflow uses strict serial processing. Each video must complete the four steps—transcription → saving → updating → triggering analysis—before the next video is processed. File names include the video ID as a reference credential, and updateContent and runAnalysisFlow always use the same ID. A transcription result from one video must never be used to update another video’s ID, ensuring that content and video records remain correctly paired.