PaperLabRead and highlight
Highlight, and the note stays put
Select a passage, pick one of five colours, write a note. Every note keeps the page and position it came from, so one click takes you back to it.
More about read and highlight →
Free and open source · runs on your computer
Read papers properly. Every note remembers its page.
PaperLab is a research reader that runs on your own machine. Highlight a passage and the note stays attached to it. Ask a paper a question and the answer shows exactly where it came from.
macOS, Windows and Linux · needs Docker Desktop · reading and notes work without any AI model
The note is the main thing, not the chat
Every tool in PaperLab ends in something you keep, with a trail back to where it came from.
Ask a paper
Answers you can check against the page
Ask a question and the answer cites the passages it used, like [C1]. Click one and the paper scrolls to that passage and flashes it. Keep the useful part as a note, marked AI.
More about ask a paper →![An AI answer citing [C1] and [C2]; hovering [C1] shows its page and section, and a line leads to that passage on the page.](/PaperLab/shots/feature-ask-a-paper-dark.webp)
Connect papers
See how your library fits together
A graph of what cites what, shared authors and topics, notes that span two papers, and papers that simply read alike. In 2D, 3D, a matrix, a timeline or rings.
More about connect papers →
Three rules it doesn't bend
Built for people who have to trust their sources, and show their supervisor where a claim came from.
Your papers stay on your machine
One user, one computer, no accounts. With a local model through Ollama, the text of your papers never leaves it. Cloud models are tagged “Cloud”, and the README lists what each setting sends and where.
How it works →AI is always labelled
What a model wrote is stored apart from what you wrote, with the model and prompt behind it. Edit an AI note and it says “AI · edited”, never “You”.
How it works →No paywall tricks
PaperLab finds free copies on arXiv, repositories and open-access publishers. When there is none, it links to the publisher and stops there.
How it works →Works with Claude
Your library, inside Claude Desktop
PaperLab ships an MCP server. Connect Claude Desktop, Claude Code or another MCP client, and it can work with your library the way you do, while everything it writes stays marked AI.
For chat inside PaperLab, pick any model per question: a local Ollama model, Anthropic with your key, or any OpenAI-compatible server.
Four tools, one library
search_librarysearches your papers and notesget_paperreads a paper’s details, outline and notesrelated_papersfinds the papers connected to onecreate_notesaves a note on a passage it quotes, marked AI
Guides for reading research well
Practical answers to what researchers ask most, whichever tools you use.
- How to read a research paper efficientlyThe three-pass method, what to skip, and how to keep what you read.
- How to take notes on research papersNotes you can find again and trace back to the passage they came from.
- How to find a free, legal PDF of a research paperarXiv, repositories, Unpaywall, CORE, and asking the authors.
- How to chat with a PDF locally, without uploading itAsk papers questions with a local model, and check every answer against the page.
- PaperLab vs ZoteroA reference manager and a reading tool do different jobs. Which one you need, or both.
- How to do a systematic literature reviewQuestion, protocol, search, two-stage screening, snowballing and PRISMA.
Who it's for
Anyone whose work depends on reading papers carefully, and showing where a claim came from.
- PhD studentsHundreds of papers, and every claim traced back to its page.
- Master’s studentsYour first literature review, read properly and cited honestly.
- Academic librariansA tool you can recommend: local, open source, no paywall tricks.
- Research groups and PIsProvenance you can check, from student notes to AI answers.
- Developers and research software engineersA full open-source RAG app to read, run and contribute to.
Who builds this
One researcher, building in the open
I'm Malek. I started PaperLab on my way to a PhD because I wanted to read papers properly, where every note stays attached to its passage and every answer can be checked against the page. It didn't exist, so I built it, and I work on it every day.
Every week the PaperLab page on LinkedIn breaks down how one feature works: how a paper becomes searchable chunks, how a cited answer is put together, how to keep AI text honest. Follow along, and tell me where it breaks on your papers.
