Exploring Your Knowledge Graph

Your knowledge base in FeynmanLM is a single, growing collection of everything you've learned — articles, papers, books, podcasts, and courses all living together. The Studio tab and Explore tab give you two different ways to navigate and interact with it.

The Studio Tab: Your Source Library

Open the Studio tab to see your knowledge base as a list of sources.

What you see for each source:

• Source type icon (color-coded: article, paper, podcast, book)
• Title and domain
• "N existing" badge in blue — how many flashcards already exist from this source
• Word count and read time estimate (once content is loaded)
• Current state (New, Ready, Reviewing, Done, Error)

Filtering and searching:

• Use the deck picker to filter sources by type (Articles, Papers, Books, Podcasts)
• Use the tag picker to filter all cards by topic across sources
• Use the search bar to find sources by title or domain

Source coverage at a glance: The "N existing" count lets you quickly identify:

• Well-covered sources (high card count) — probably don't need more cards
• Under-covered sources (low or zero count) — good candidates for a new generation pass
• New sources — just added, cards not yet generated

The Explore Tab: Visual Concept Explanations

The Explore tab takes a flashcard concept and generates an interactive visual explanation — a React-based applet with diagrams, step-by-step walkthroughs, and interactive controls rendered directly in the app.

How to use it:

1. Open the Explore tab.
2. Enter a concept (you can type a topic, paste a card question, or select from your card collection).
3. Tap Explore — the AI generates a visual explanation tailored to that concept.
4. Interact with the rendered applet: adjust sliders, step through a diagram, or explore alternative representations.
5. Previous explorations are saved in the history sidebar so you can revisit them.

What kinds of concepts work well:

• Multi-step processes (metabolic pathways, algorithm steps, protocol flows)
• Systems with multiple interacting components (neural network layers, distributed systems)
• Mathematical concepts with visual geometry (linear transformations, probability distributions)
• Historical timelines or causal chains

Tags as a Knowledge Map

Tags are your primary organizational tool across the entire knowledge base. Every flashcard can have multiple tags, and you can filter your review, Studio, and searches by tag.

A healthy tag structure might look like:

machine-learning
  ├── transformers
  ├── optimization
  └── generalization
biology
  ├── cellular-respiration
  └── genetics
cs
  ├── distributed-systems
  └── algorithms

Viewing all tags: Tap All Tags on the Setup screen to open the tag picker. It shows every tag in your collection with a count of how many cards have that tag. Tags with higher counts represent areas of dense coverage.

Decks as a Topic Hierarchy

The five main decks reflect source type, not topic:

• Articles — from web articles
• Papers — from academic papers
• Books — from book chapters
• Podcasts — from podcast transcripts
• Courses — from course materials

Sub-decks let you go deeper. For example, Papers > Machine Learning groups all ML papers together.

Deck picker: tap All Decks on the Setup screen to select a specific deck for review.

Cross-Source Connections

One of FeynmanLM's key advantages over a tool like NotebookLM is that all sources live in the same space. When you review a card about transformer attention, the AI can draw on the source context from whichever paper or article that card came from — while your entire card history provides the background of what you already know.

This means the follow-up questions during review are calibrated to your actual knowledge state, not just the isolated content of one notebook.

Generating Distill-Style Explainer Articles

For topics where you have deep flashcard coverage, the app (and MCP) can generate a Distill-style interactive article:

Via Claude (with MCP connected):

"Draft an explainer article on my oxidative-phosphorylation tag cluster"

Via CLI:

uv run python scripts/article_draft.py --topic "oxidative-phosphorylation"
uv run python scripts/article_build.py --slug oxidative-phosphorylation

These articles compile your accumulated knowledge into a published, interactive HTML document — a tangible artifact of what you've learned.