Papers

Academic papers — from arxiv, Nature, Science, bioRxiv, and similar journals — get special treatment in FeynmanLM. The app auto-detects them by URL, routes accepted cards to the Papers deck, and automatically fetches or locates the full PDF whenever possible.

How Papers Are Detected

FeynmanLM classifies a source as a paper when its URL matches known academic domains:

• arxiv.org
• nature.com
• science.org
• cell.com
• biorxiv.org
• medrxiv.org
• plos.org
• pubmed.ncbi.nlm.nih.gov
• and other major journal publishers

Adding a Paper

From Safari Reading List: Save the paper's abstract or landing page to Safari Reading List. The app picks it up automatically within the next session (articles added in the last 7 days are surfaced).

Manual paste: Open the Studio tab, paste the paper URL into the source bar, and press Return.

Browser extension: While reading a paper online, click the extension icon or right-click → Capture to FeynmanLM.

How PDF Loading Works

When you add a paper, FeynmanLM tries to get the full PDF automatically and store it in your iCloud. The resolution order is:

1. Local iCloud PDF — checks if a PDF already exists in your FeynmanLM/Papers/ folder in iCloud Drive (from a previous fetch or manual download)
2. Direct fetch — tries to download the PDF from the URL (works for open-access papers on arxiv, bioRxiv, PLOS, etc.)
3. PubMed Central (PMC) — if the paper is available in PMC, the app fetches the full text automatically
4. Nature (with MagicWord) — if you have a Nature MagicWord institutional login configured, the app can access Nature papers automatically (see below)
5. Skip — if none of the above work, the paper shows an error state with a Retry button

When a PDF is fetched successfully, it's saved to your iCloud Drive so it's available on all your devices:

~/Library/Mobile Documents/com~apple~CloudDocs/Documents/FeynmanLM/Papers/

In Finder: iCloud Drive → Documents → FeynmanLM → Papers.

Dealing with Paywalled Papers

Many papers are behind paywalls. Here's how to handle them depending on the publisher:

Open access (no action needed)

Papers on arxiv, bioRxiv, medrxiv, PLOS, and other open-access repositories are fetched automatically. Nothing to configure.

PubMed Central (PMC)

If a paper is available in PMC (even if the publisher's version is paywalled), FeynmanLM detects this and fetches the PMC version automatically. Many NIH-funded papers end up in PMC 6–12 months after publication.

Nature (MagicWord login)

If you have institutional access to Nature via a MagicWord login, you can configure it in the app:

1. Go to Settings → Nature Access.
2. Enter your MagicWord credentials.
3. FeynmanLM will use these to access Nature, Nature Methods, Nature Biotechnology, and other Nature Portfolio journals automatically.

Other paywalled publishers

For papers behind paywalls that aren't covered by the above (Science, Cell, Elsevier, etc.), you'll need to get the PDF yourself:

1. Download the PDF from your institutional access or legal preprint repository.
2. Save it to the FeynmanLM/Papers/ folder in iCloud Drive.
3. Rename the file clearly — the filename helps the app match it to the URL:

   attention-is-all-you-need.pdf
   vaswani-2017-transformer.pdf

4. Next time you select the paper in Studio, the app finds the local PDF automatically.

The app uses fuzzy filename matching, so exact filenames aren't required — though cleaner filenames help.

Generating Flashcards

1. Select the paper in the Studio sidebar.
2. Click Generate Flashcards.
3. The AI focuses on:
   • The core hypothesis and findings
   • Key mechanisms and causal chains
   • Methodology nuances (especially for papers where how the result was obtained matters)
   • Surprising or counterintuitive findings

For papers, the Make Harder refine option is especially useful — it shifts questions from "what did they find?" to "why does this mechanism work, and what would change if assumption X were wrong?"

Context Files for Deep Review

For deep paper review — where the AI should quote the paper when evaluating your answers — upload the PDF as a Context File:

1. Tap the book icon in the top toolbar to open Context Files.
2. Tap + and select the PDF from iCloud Drive.
3. During card review, toggle Book context on the question screen to include the paper text in the AI's evaluation context.

This is especially useful for dense technical papers where you want the AI to cite specific passages when giving feedback.

Tips

• Preprints are fine — arxiv and bioRxiv papers work especially well since they're always publicly accessible.
• Long papers: the app reads up to 100 pages / 50,000 characters. For very long papers, the AI focuses on the abstract, introduction, and conclusion sections if the full text is truncated.
• Scanned PDFs: PDFKit performs basic OCR on scanned documents, but the quality is lower than machine-readable PDFs. Always prefer a native PDF over a scan.
• Papers with lots of figures: the text extraction skips images, so cards will be text-based. Use the Generate Diagram feature during review to create visual explanations for figure-heavy content.