arXiv
Subscription Viewer

Find your next paper.
Keep your place.

Review arXiv papers from your terminal. Scan abstracts, star promising work, and keep a reading queue with local notes and citation exports.

Start your first review

Open source · Python 3.13+ · No AI service required

Terminal browser with papers on the left and the selected abstract on the right. The footer shows search, selection, read state, and export shortcuts.
One place to scan, decide, and return.Development capture · fictional papers · View full size

Your first papers,
in two commands.

Start with live search. You need an internet connection and a terminal; no account, API key, or saved email is required.

Already keep an archive? Browse local history.

1. Install the published package

Use uv to keep the tool in its own environment. It can provision the required Python 3.13.

uv tool install --python 3.13 arxiv-subscription-viewer

2. Load a small page of recent papers

arxiv-viewer search --category cs.AI --mode page --max-results 20

Dismiss the welcome with Enter. You should see up to 20 papers and the selected abstract. Replace cs.AI with your field.

Need pip or help with your shell? Installation alternatives · First-run recovery

From a long list to a short reading queue.

A useful session starts with your own decisions. Research services can wait.

  1. Scan the abstracts

    Move with j / k or the arrow keys. Read the selected paper beside the list.

  2. Keep what matters

    Press x to star, n to add notes, and r to mark a paper read. Your decisions persist.

  3. Take it into your work

    Select with Space, then E to export BibTeX, RIS, or CSV files, or copy Markdown.

Filter loaded papers with / and starred. Press ? for help or Ctrl+p for commands. Keep the cheat sheet nearby.

Your daily emails
have a place here, too.

Save arXiv subscription emails as dated text files in a history/ folder. Launch from its parent directory, move between dates with [ and ], and return to your saved review state.

Set up a local history archive 

research/
└── history/
    ├── 2026-01-21.txt
    └── 2026-01-23.txt

From research/:

arxiv-viewer browse

Plain-text arXiv email bodies, not generated Markdown. First use opens the newest file; later runs resume the saved date.

Go further,
one task at a time.

Focused guides for the next step in your reading workflow.

Guides open as readable Markdown on GitHub and track development. Check arxiv-viewer keybindings for your installed version.

All documentation
Find and filter papersLive queries, local filters, and saved searches Build a reading list or bibliographyCollections, citations, PDFs, and metadata backup Make a daily or weekly briefMarkdown digests and an interactive inbox Make it fit your workflowThemes, storage paths, and optional services Get unstuckInstallation, missing papers, and terminal behavior

Local notes. Optional outside context.

The core workflow is yours

Stars, notes, tags, collections, and session state are stored locally. Run arxiv-viewer config-path to find the configuration file. Citation exports and metadata backups let you carry your work elsewhere.

Saved email text works offline. Live search, missing abstracts, PDF downloads, and figure retrieval need the network.

Add research services when useful

Semantic Scholar adds citation context; Hugging Face adds daily-paper signals. Both are disabled by default.

AI summaries and chat require your own configured CLI or HTTP provider. Paper content and prompts go to that provider; a local CLI may still call a hosted service.