AutoWeft

AutoWeft is an AI assistant in your browser sidebar. Ask questions about the current page, organize its content into tables or notes, and transcribe supported videos. Configure a chat model before use; video transcription can use local or cloud speech recognition.

This 61-second recording shows a Bilibili video being transcribed. After a request in the sidebar and approval to extract the audio, a local Whisper model produces a timestamped transcript.

Setup after install

Configure a chat model in Settings. To transcribe videos, also select a speech recognition service or download a local model.

1 · Configure the chat model

Open Settings → Models in the sidebar.

  1. Select a provider preset or enter a service URL compatible with the OpenAI or Anthropic API.
  2. Enter the API key and model name. The key is stored on your device and used to authenticate requests to the configured service.
  3. For a local or self-hosted service such as Ollama or vLLM, enter its compatible API URL and choose a model that supports tool calls.

2 · Speech recognition (ASR)

In Settings → Models, find Speech recognition (ASR).

  • For cloud recognition, select a provider and enter its API key. Audio is sent to that service, which may charge for use.
  • For local recognition, select Local model and download Whisper. The tiny, base and small models require approximately 40, 80 and 250 MB respectively.
  • Local speech recognition runs in the browser without a speech service API key or an audio upload to a cloud transcription service. If you later process the transcript with a cloud chat model, the relevant text is sent to that model.

After setup, enter “transcribe this video” in the sidebar. Audio extraction depends on the page’s media format and access restrictions. For more setup instructions, see the GitHub README

What it can do

Describe what you need in the sidebar. AutoWeft attempts to read the current page and use the relevant tools to process its content.

💬 Page Q&A & analysis

Ask questions, summarize articles or compare information using the page content that AutoWeft can read. Answers can contain errors; check important details against the source.

📊 Data extraction & processing

Save page content as tables or Markdown notes. You can view, search and download saved results from the Data tab.

🎬 Video transcription

Extract audio from supported video pages and generate timestamped transcripts with your selected speech recognition model. Encryption, login requirements and site restrictions can prevent extraction.

📄 Full-page export & archive

Export readable page content as Markdown and attempt to save associated images. You can view and download the result from the Data tab.

How your data is handled

Conversation history and saved materials are stored on your device. When you use a cloud model, the page content or transcript needed for the request is sent to your configured service. Cloud speech recognition also sends audio to that service.

AutoWeft does not require a product account. Your model provider may require an account and API key. Anonymous usage statistics are enabled by default and can be disabled in Settings.

Operations that require authorization show the tool and target. You can approve them individually or enable automatic approval for the current session.

Read the privacy notice

Install / Update

Install from this page by choosing a local folder. The installer downloads and writes the extension files, which you then load from your browser’s extensions page.

Latest:loading…   Installed:unknown

. Routine updates skip unchanged files. Use a full check if you suspect that installation files are damaged.

Installation size and file contents

The current download is about 20 MB, or 65 MB unpacked. The main files are:

  • Audio processing engine: about 32.2 MB, for extracting audio tracks and converting formats.
  • Local speech recognition runtime: about 26.9 MB, for running recognition in your browser.
  • PDF character maps: 168 files totaling about 1.2 MB, for text extraction including CJK.
  • Other files: about 4.3 MB of interface, task logic, PDF parsing, fonts and licenses.

Sizes above are unpacked. Speech model weights are not included; download them separately in extension settings when needed. After unpacking, you can inspect every file’s actual size and purpose.


          

            First-install guide
            
    
              
  1. Click ① to create or pick an empty folder (e.g. autoweft-ext).
    2. 
              
  2. Click ② and wait for the files to be written.
    3. 
              
  3. Click ③ and follow the dialog to load or refresh on chrome://extensions.
    4. 
              
  4. For updates, select the original installation folder, write the files, then reload the extension from your browser’s extensions page.
    5. 
            

            
Chrome / Edge only (requires the File System Access API).

          

        

      


      
      
        

          
③ Reload the extension

          
Web pages cannot navigate to chrome:// URLs. Copy the address below and open it in the address bar:

          

            chrome://extensions/
            
          

          

            
First install (not loaded yet)

            
    
              
  1. Turn on Developer mode (top right of the extensions page).
    2. 
              
  2. Click "Load unpacked" and pick the folder from step ①.
    3. 
            

          

          

            
Update (already loaded)

            
    
              
  1. Find the extension you loaded on the extensions page.
    2. 
              
  2. Once the files have been written, click the extension’s Reload button.
    3.