Memory Cache
Basic Information
Memory Cache is a small project that captures web pages from Firefox as PDF files and places them into a synchronized 'MemoryCache' folder so those documents can be consumed by a local retrieval workflow built around privateGPT. The repository includes a Firefox extension that adds a toolbar button to save the current page, instructions to load the extension temporarily in Firefox, guidance to patch Firefox to enable printerSettings.silentMode for automatic PDF saving, and a script-based ingest workflow that watches the downloads folder and moves new PDFs into privateGPT's source_documents directory. The README documents prerequisites, setup steps for the extension, and how to connect the saved PDFs into a privateGPT instance.
Links
Stars
564
Github Repository
Categorization
App Details
Features
A lightweight browser extension to save the active page as a PDF from Firefox. Instructions to load the extension temporarily via about:debugging and add the extension to the toolbar. Guidance for patching Firefox to add a printerSettings.silentMode property so saving can occur silently. Use of a MemoryCache directory that is symlinked into privateGPT"s source_documents location to centralize saved documents. A provided run_ingest.sh script that uses inotifywait to watch the downloads folder and trigger ingestion into privateGPT. README includes setup prerequisites and stepwise usage instructions to integrate saved PDFs with a local LLM ingestion pipeline.
Use Cases
The project streamlines capturing web content for on-device retrieval-augmented workflows by automating the conversion of pages to PDF and placing them where a local privateGPT instance can index them. It reduces manual steps for researchers or users running privateGPT by providing clear setup steps, a temporary extension load path for Firefox, and an automated watcher script to move new items into the model"s source_documents directory. By enabling silent PDF saving with a patched Firefox property and a synchronized MemoryCache folder, the repository helps maintain a local corpus of web snapshots to augment private LLM responses without relying on remote services.
-
$$$$ -
$$$$