Users are bypassing Amazon's ecosystem to link Kindle libraries with Claude AI
By jailbreaking Kindles and using open-source tools, readers can now use Claude to analyze their personal book collections for tailored recommendations.
The short version
- Amazon discontinued the 'Download & Transfer via USB' feature for Kindle ebooks in February 2025.
- Users can use OpenMTP to move book files from a Kindle device to a computer for local management.
- The KOAssistant plugin allows Claude to analyze a local folder of EPUB files to provide personalized reading suggestions.
- The setup requires a jailbroken Kindle, a Calibre-Web server, and an API key from an AI provider like Anthropic.
Why it matters
This workflow shifts control from Amazon's proprietary algorithms to a user-owned local library. It allows for a more personalized discovery experience that understands a reader's specific tastes rather than just popularity metrics.
Breaking free from Amazon's ecosystem
Amazon has made it increasingly difficult for users to manage their ebook libraries outside of its own software. The company officially discontinued the 'Download & Transfer via USB' feature for Kindle ebooks in February 2025, which limited how people could move files onto their devices. This move makes it harder for users to maintain independent backups or use third-party management tools.
To circumvent these limitations, some users are turning to jailbreaking their devices. By unlocking the hardware, readers can install open-source software like KOReader. This app serves as a document viewer and ebook reader that supports a wider variety of file formats than the standard Amazon software.
This transition allows the Kindle to function as a more flexible device. Instead of being tied to Amazon's specific ecosystem, the hardware can be used to read files hosted on personal servers or local folders. This shift is driven by a desire for better library management and more control over personal data.
Building a personal library server
The first step in this new workflow involves moving books off the Kindle and onto a personal network. Since macOS no longer mounts Kindles as plain drives over USB, users can use a tool called OpenMTP to copy files from the device's documents folder. This process allows for a complete backup of a user's entire collection onto a computer.
Once the files are on a computer, they often need to be converted. Tools like Calibre can turn .azw3 files into the more universal EPUB format, though this does not work for files with DRM protection. For those books, users may need to use additional plugins like DeDRM to remove digital rights management before hosting the files.
To make the library accessible, users can set up a Calibre-Web instance on a home server. This creates a local web feed where books can be browsed and downloaded to any compatible device. It provides a centralized hub where every title is tagged with specific metadata, making it much easier to organize than a standard folder of files.
Integrating Claude AI for discovery
The most advanced part of this setup involves connecting the library to an AI assistant. By using the KOAssistant plugin within KOReader, users can link their personal library to Claude. This requires an API key from Anthropic, which is billed per use rather than through a standard chat subscription. The plugin allows the AI to understand the user's specific reading history and preferences.
To keep the process organized, users are encouraged to put their EPUB collection into a single dedicated folder. When the KOAssistant performs a library scan, it only sees the titles and authors in that specific folder. This prevents the AI from seeing unrelated files while still giving it enough information to recognize patterns in the user's taste.
The AI can then provide insights that Amazon's popularity sorting cannot match. For example, if a user has a collection of crime and mystery novels, Claude can identify those themes and suggest specific authors. Users can also ask the AI to compare a book they are currently reading with others in their library to find similar themes.
Technical requirements and limitations
Setting up this system requires several specific technical steps and hardware permissions. Users must have a jailbroken Kindle Paperwhite (12th generation) or a similar compatible device to run KOReader. They also need a stable Wi-Fi connection and a server to host the Calibre-Web instance for the library to function correctly.
The process involves manual configuration of files like apikeys.lua to input the necessary credentials. Because the AI's access is limited to the scanned folder, users must be intentional about what they include in the scan. This ensures that the AI only receives the metadata it needs to provide recommendations without accessing the full text of every book.
While the setup is complex, it offers a level of customization that is unavailable in the standard Kindle app. It moves the responsibility of book discovery from a corporate algorithm to a personal AI assistant. This allows for a more nuanced experience where the AI acts as a personal librarian who knows exactly what the reader enjoys.
What comes next for independent readers
This trend highlights a growing movement toward 'de-Googling' or 'de-Amazoning' personal media libraries. As companies restrict how users can interact with their own purchased content, more people are seeking open-source alternatives. This setup provides a blueprint for how readers can maintain ownership of their digital collections while still using modern AI tools.
The next step for many users is to refine how the AI interacts with their unread books. Some are planning to have the AI scan folders of books they haven't started yet to provide a prioritized reading list. This turns a static collection of files into a dynamic, interactive experience that adapts to the reader's current mood.
As AI models become more capable of understanding nuance, these personal library integrations will likely become more sophisticated. Readers may soon see more complex queries, such as asking for books that match the specific emotional tone of a favorite novel. For now, it remains a manual but powerful way to reclaim the reading experience from proprietary software.
How it works
The process begins by extracting files from the Kindle device using OpenMTP, a tool that bridges the gap between the device and a computer. Once the files are on a Mac or PC, they are organized and converted into EPUB format using Calibre. This ensures the files are compatible with various open-source readers and can be managed with detailed metadata tags.
Next, the user hosts these files on a Calibre-Web server, which acts as a private cloud for their ebooks. This allows them to browse their library from any device that supports an OPDS feed. By moving the files off Amazon's servers, the user ensures that their library remains accessible even if Amazon changes its software policies or removes specific titles.
Finally, the KOAssistant plugin connects the local library to Claude via an API key. When a user performs a 'Library Scan,' the plugin sends the list of titles and authors to the AI. Claude then processes this list to provide tailored suggestions, comparisons, and genre analysis based entirely on the user's personal collection of books.
Who it affects
This workflow is primarily for power users who are frustrated by Amazon's restrictive ecosystem. It appeals to readers who want to own their files and move away from the 'walled garden' of the Kindle app. These users often value privacy and want to ensure their reading habits aren't being tracked by a single corporation.
It also affects the way people discover new content. Instead of relying on 'Best Seller' lists or sponsored ads, these readers use a personal AI to filter their own archives. This creates a more curated experience where the recommendations are based on the specific nuances of a person's past reading rather than general trends.
While it requires a bit of technical effort to set up, it offers a high degree of autonomy. It is a solution for the curious reader who wants to use the latest AI technology to enhance their hobby without sacrificing control over their personal data or the way they access their books.
What comes next
The most immediate next step for users is to expand the AI's role in the reading experience. Some are already looking to have Claude scan folders of unread books to generate a 'what to read next' list. This would move the AI from a simple search tool to a proactive recommendation engine that understands the user's backlog.
There is also potential for more complex interactions, such as asking the AI to find books with specific themes that the user hasn't explored yet. As the KOAssistant plugin and other open-source tools evolve, the way we interact with our personal libraries will become increasingly conversational and personalized.
Finally, this trend may encourage more people to explore open-source ebook readers and self-hosted servers. As more people find success with these setups, it could lead to a wider adoption of independent reading tools that prioritize user ownership over corporate convenience.
Why it matters
This setup changes the power dynamic between the reader and the platform. By moving the library off Amazon's servers, users reclaim the ability to manage their own data and choose how they interact with it. It turns a passive consumption model into an active, managed experience.
For the reader, it means a more personalized discovery process. Instead of being fed what is popular, they are shown what is relevant to them based on their own history. It allows the AI to act as a private librarian, providing a level of curation that a generic store algorithm simply cannot achieve.
Ultimately, it demonstrates how open-source tools can be combined to bypass corporate restrictions. It shows that with a bit of technical setup, users can still enjoy modern conveniences like AI while maintaining independence from the software companies that provide the hardware.
Try it
Check if your Kindle can be connected to a computer using OpenMTP to see if you can move your current book files to a local folder.
Sources and links
- CodeKOReader app github.com
- Codegithub.com/ganeshrvel/openmtp github.com
- Reportingxda-developers.com: the article we sourced
Read the original at xda-developers.com
First reported by xda-developers.com. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.