# Literature notes, a paper per note Each paper is a note: the PDF, the two quotes that matter with their pages, and what you think. When your supervisor asks who disagrees with Ahmed, a tag finds them. ## At a glance - **For:** PhD students, researchers and anyone writing a literature review. - **You keep:** A thread per research question, and a note per paper. - **Nothing to set up:** Each paper as a note: the PDF, the quotes that matter, and what you think. - **The moment:** Your supervisor asks which papers disagree with Ahmed et al. A search for #disagrees lists them, each with its quote and page. - **Plan:** Free · Free keeps your notes in this browser; export a backup now and then. Plus adds Sync across your own devices. Pro with Sync adds sharing and the optional connections below. ## Before The papers, quotations and findings in the example below are fictional. Forty papers for the review chapter. The PDFs sit in a downloads folder named by DOI, the highlights inside each PDF, and your thoughts in a document you stopped updating in May. Your supervisor asks which papers disagree with Ahmed et al., and you open them one by one. ## Why a thread, not a folder of PDFs ### A paper is a note The PDF, the citation, the two quotes that matter with their pages, and your own line about it. ### Papers link to papers A Note reference from one paper to another says which one it argues with. ### Your own tags #method, #disagrees, #dataset: Search lists every paper with a tag, across questions. ## A thread, over time Your workspace starts empty. ### Remote work and sleep Literature review · Active #### Thu 3 Sept *A paper is a note: the PDF, the quote and its page.* **You** · 10:00 Ahmed et al. 2024, the study everyone cites: “remote workers sleep 24 minutes longer” (p. 6). #baseline [Image: PDF of Ahmed et al. 2024] #### Sat 5 Sept *A Note reference to the paper it argues with.* **You** · 16:20 Novak and Ruiz 2025: no difference in sleep once the commute is taken into account (p. 11). One country only. #disagrees ↳ Ahmed et al. 2024 · 3 Sept #### Fri 18 Sept **You** · 11:10 Chen 2023: the same direction as Ahmed, with a smaller sample. #agrees #### Wed 14 Oct *With Pro, your supervisor writes in the same thread.* **Dr. Mensah** · 15:30 Add Sato’s 2025 replication before you claim it. #### Fri 16 Oct *The paper your supervisor asked for, a note like the others.* **You** · 09:40 Sato 2025, a replication in three countries: 19 minutes longer (p. 4). #agrees #### Tue 20 Oct *With Pro, Claude reads the thread through its app and adds the map as a note.* via **Claude** · 17:05 The papers in this thread, by what they find: | Finding | Papers | |---|---| | Remote workers sleep longer | Ahmed 2024, Chen 2023, Sato 2025 | | No difference once the commute is counted | Novak and Ruiz 2025 | ## If you like, let it file itself. None of this is needed to start. It takes a few minutes once, and needs Pro. [Script or API](https://thread.page/help/connections/custom.md) ### A map of the thread, from Claude or ChatGPT Ask the app to read the thread and add a table of who finds what, with the papers behind each line. #### Set it up App Read and add Selected folders and threads: Literature review Pro with Sync 1. In **Settings → Connections → Apps**, choose **Connect an app** → **Claude app** or **ChatGPT app** and follow its steps. On ThreadPage’s consent page, choose **Customize access** → **Selected folders and threads** → **Literature review**. 2. Ask: “Read Remote work and sleep and add a table of what the papers find, with the papers behind each finding.” ### Ask the thread Or write to @Claude in the thread, on your own API key. It answers with a note. #### Set it up AI model Your Anthropic or OpenAI API key Pro with Sync 1. In **Settings → Connections**, open **AI models**, choose **Claude** or **GPT**, paste an API key from Anthropic or OpenAI and choose **Connect**. AI models need Pro with Sync turned on. 2. In the thread, write `@Claude` and your question, then choose **Send**. In a thread you share, choose **Allow here** the first time. ### New papers from arXiv A workflow in n8n or Make watches an arXiv search and adds each new paper to a To read thread, with its title, authors and link. #### Set it up Automation Add notes from another app Text or Markdown Pro with Sync 1. In **Settings → Connections → Automations**, choose **Connect an automation** → **Add notes from another app**. Name it arXiv, choose your To read thread and **Text or Markdown**, and copy its URL. 2. In n8n or Make, start from an RSS trigger on the arXiv search you follow, and add an HTTP request to the URL that sends `{"id": "…", "markdown": "…"}` with the title, the authors and the link. ## Also Writing the chapter with Claude Code or Codex? With Pro, connect it with **Read only** access to Literature review (see [Coding agents help](https://thread.page/help/connections/coding-agents.md)), and it can read the papers’ notes while it drafts. ## Limits - There’s no citation manager, PDF annotation or equations: keep your reference manager for citations. - Search finds a PDF by its file name, not the text inside it: copy the quotes you need into the note. - On Free your notes stay in this browser: export a backup now and then. Claude, ChatGPT, Claude Code, Codex, Anthropic, OpenAI, arXiv, n8n, and Make are the names of their owners. ThreadPage is independent and not affiliated with or endorsed by them. ## Start with one thread. [Start writing](https://thread.page/app) Free in your browser. No account needed to start. More use cases ### [A lab notebook your script fills in](https://thread.page/use-cases/lab-notebook.md) Every run arrives with its parameters, table and plot. Your chapters link to the run behind each claim. Pro · results from a script ### [A research report you’ll need again](https://thread.page/use-cases/research-report.md) ChatGPT or Claude can save a report alongside your own notes. Keep the decisions and follow-ups with it, then ask a connected model about the thread when you need to. Pro · connected apps; your key for @ replies [All use cases](https://thread.page/use-cases.md)