Learn from materials skill
learn-from-materials
A skill that turns books, PDFs, slides and documents into a traceable knowledge base and an interactive learning page with sources, maps and self-check
- 1,100 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Runs local Python scripts that read your books and documents and create a knowledge base and HTML pages beside them
- May call installed parsing programs (OCR, Calibre) when processing scans and e-books
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/learn-from-materialsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Runs local Python scripts that read your books and documents and create a knowledge base and HTML pages beside them
- May call installed parsing programs (OCR, Calibre) when processing scans and e-books
This is third-party code. Review the repository files before installing.
What it does
Learn from materials breaks learning material down in two stages: scripts first extract the text and build a layer of facts, then the agent writes the content as JSON and a fixed renderer builds the page from it. The skill does not hand-write HTML. It accepts PDF, EPUB, MOBI, AZW, AZW3, DOCX, PPTX, PPTM, HTML, Markdown, TXT, RTF and sets of several files. There are two depths. A quick overview scans the whole structure and gives the core arguments and limits. Systematic study reads each source block, keeps a coverage ledger, checks the summary claims back against the material and verifies the source file hashes. The output is three results from the same data: a `.learnkb/` knowledge base, one HTML file with no external dependencies and a Markdown version. The page has six modules: a framework of ideas with a relationship map, guided content, a glossary, action rules, dynamic self-check and local notes. The skill keeps a method library in `methods.json` and `patterns.md`: cards with purpose, prerequisites, steps, an effect check and a short quotation that can be searched and compared across materials. The «Apply this methodology» button assembles a prompt for the agent with your question and constraints.
Who it is for. For students, teachers and anyone studying books, papers and slide decks who wants to see which page of the source each conclusion rests on.
Good fit when
- You want to work through a book or a PDF report and get a digest where the core ideas carry a source page
- You need a quick overview of a slide deck tied to slide numbers and an offline page that opens without a network
- You want to build a method library from several materials and compare the authors' approaches
- You want self-check on the material: questions, notes and a mistake log that stay only in the browser
- You want to apply a method from a book to your own task and get a prompt with conditions and limits
Not a fit when
- You need a Russian-language result: the skill supports English and Chinese, and for another language it states the limit and asks you to pick one of the two
- Your agent cannot read files or run Python: without a file system only part of the prompts works, with no extraction, coverage checks or page build
- A file is scanned without a text layer and the agent has no OCR or vision: the skill marks those ranges as unverified
- You need a digest of a MOBI, AZW or AZW3 file without Calibre installed: these formats have no fallback parser
- You only need a paragraph or two of summary: the full pipeline with its check ledgers is built for long materials
Example request
Use learn-from-materials to systematically study this PDF and build a traceable knowledge base and a learning pageLimitations
- Result languages: English and Chinese (zh-CN). The language is chosen by an explicit instruction, then by your request, not by the material's language or browser settings.
- It needs Python 3.10 or newer and an agent with file access. The skill does not install extra libraries (`pdftotext`, `Docling`, `OCRmyPDF`, `ebooklib`, `python-docx`, `striprtf`) itself; without them it takes the next parsing path or marks the range unverified. MOBI, AZW and AZW3 need Calibre.
- The checks verify the mapping to the source but do not prove semantic truth or the absence of omissions. The first output ships as static-only, and the author asks for browser visual review to be run separately.
- Long books and cross-material analysis work better on models with a long context and strong reasoning.
- MIT licence: the project builds on `book-to-skill` and `book-to-webpage`, whose authors did not endorse the extensions. Process only material you have rights to and do not publish knowledge bases containing someone else's text.
How to disable. Delete the `learn-from-materials` folder from the agent's skills directory (`~/.claude/skills/` for Claude Code, `~/.agents/skills/` for Codex) and restart the client. Generated `.learnkb/` bases and pages sit next to the materials; delete them by hand. Notes and self-check answers live in the browser's local storage; clear them in browser settings.
FAQ
How do I make a digest of a PDF or book with source pages?
Ask the agent to use the skill and pick a depth. For PDFs and slides the skill maps a conclusion to the page or slide number, and for EPUB to the chapter. When the exact place cannot be confirmed, the page shows a block-level or conclusion-level range instead of an invented number.
How do the quick overview and systematic study differ?
A quick overview scans the structure and checks the displayed citations; no exhaustive question bank is built in advance. Systematic study reads each block, keeps coverage and rule ledgers, runs an independent second content review and indexes questions. If you do not name a depth, the skill asks once.
Does the skill send my materials online?
The project's scripts do not send materials over the network, and browser verification lets through only the HTML itself and `data:` and `blob:` resources. What you hand to the agent is processed under your platform's rules and account settings. Notes and the reader profile stay in the browser.
Which file formats work without extra programs?
Most formats are parsed with the Python standard library or system tools. Text PDFs have a chain of fallback parsers, and EPUB and DOCX fall back to archive parsing. Only MOBI, AZW and AZW3 do not open without a separate install: they need Calibre `ebook-convert`.
What is the method library and how do I use it?
After parsing, the skill saves the author's techniques into `methods.json` and builds a readable `patterns.md`. A card stores purpose, prerequisites, limits, steps, an effect check, a short quotation, a stable ID and a version. You can index several method files, search candidates by Chinese or English keywords and export comparison records.
What does the «Apply this methodology» button do?
You enter a question, goal and constraints, choose the whole structure, one method or automatic selection, and the page assembles a prompt. You copy it into the conversation with your agent. The page runs no model itself: the prompt asks the agent to assess applicability first and then give a sourced analysis.
Discussion
No questions or reviews yet.
Related
SkillsA skill that typesets resumes, one-pagers, letters, reports and slides from templates into PDF, PNG and PPTX and checks the result before delivery
Turns a technical book or a folder of documents into a structured skill with chapters, a glossary and a cheatsheet
A skill for writing and revising ML, CV and NLP papers: section structure, paragraph flow and reviewer-style self-review
A skill that translates whole books in PDF, DOCX and EPUB into any language using parallel subagents
The paper-spine skill takes a topic, materials and data through a scientific paper: literature, text, figures, review and Word, LaTeX and PDF files
Skills tailored to specific academic journals that help choose a venue, prepare a paper and pass review
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| Learn from materials (this entry) | no | 1,100 | |
| tutor-skills | no | 1,358 | turns PDFs, documents and a codebase into an Obsidian study vault with structured notes |
| notebooklm-skill | no | 480 | works through Google NotebookLM: notebooks from links and files, cited answers, podcasts and quizzes |
README in short
The description in English and Chinese opens with features: a methodology for the whole material, a cumulative method library, page language and interactivity, and learning depths. Sixteen page screenshots follow for a biography, a market report and a research paper. Then come install steps per agent, example requests, requirements, a table of optional components, commands, security and acknowledgements. The repository holds `SKILL.md`, the `references` folder, `scripts` with tests and `examples`, a changelog and a security document.
README badge
Original SKILL.md
--- name: learn-from-materials description: Turn books, PDFs, slides, documents, web pages and text into a source-grounded knowledge base, interactive learning HTML and Markdown in English or Chinese. Use for learning, summaries, explanations, review, quizzes, relationship maps, saving reusable methodology files, retrieving or comparing accumulated methods, and applying them to a real problem. license: MIT; see LICENSE.md and NOTICE.md metadata: version: "1.0" revision: release compatibility: Requires filesystem access and Python 3.10+. Node.js and Playwright are optional for browser verification. Visual or OCR capability is recommended for scanned PDFs and image-heavy slides. --- ## Language routing — before your first reply Resolve output language in this order: explicit user preference → language of the current substantive user request (ignore attached material, quotations, code and paths) → established conversation language. English requests produce English; Chinese requests produce Chinese. If genuinely ambiguous, ask one short language question together with any required learning-depth choice. Do not infer nationality or language from identity, browser settings, filenames, or the material's language. This version supports `en` and `zh-CN`; for another requested language, state the limit and ask which supported language to use.