Skip to main content
litman

Website PyPI version Python 3.12+ License: MIT AI-native: agentic tool

Keep every paper you read on your own computer, and let an AI assistant do the filing.

Open a paper, highlight it, write your notes beside it — everything stays in ordinary files and folders that you can read, copy, and back up yourself, with or without litman. For anything past reading, just say what you want — add this paper, which ones did I read for the peptide project? — and your AI assistant does it.


Install

macOS / Linux:

curl -LsSf https://raw.githubusercontent.com/wqx1999/litman/main/install.sh | sh

Windows (PowerShell):

powershell -ExecutionPolicy ByPass -c "irm https://raw.githubusercontent.com/wqx1999/litman/main/install.ps1 | iex"

Double-click the litman icon the installer creates — desktop on Windows, Launchpad on macOS, applications menu on Linux.

Install with pipx

pipx builds litman's environment from the Python already on your machine, so it needs Python 3.12 or newer on your PATH.

pipx install litman
lit gui --make-shortcut    # the icon — the one-line installer does this step for you
Install from a local clone (development)
# first install
git clone https://github.com/wqx1999/litman.git
cd litman
pipx install .

# update (pull latest code first)
git pull
pipx install --force .

The AI assistant

Reading and annotating work on their own. The just say what you want layer runs on an AI agent, which you install separately — litman ships the skills it drives, not a model. lit agent starts yours in the library directory; the app's agent button does the same.

Claude Code is the supported agent today. Codex, Cursor, Gemini CLI, and OpenCode are next: they already appear in litman's agent picker, greyed out until the release that turns them on.

Your agent works with whatever model you give it. We ran eight of them through 22 everyday tasks:

  • Recommended: an Anthropic subscription. Claude Sonnet 4.6 and Haiku 4.5 both complete 97% of the tasks.
  • Without one, DeepSeek-V4 scores highest of the rest.

Every write is validated whatever the model, so a weaker one needs more turns but never corrupts the library. Method and per-model scores: agent model benchmark.

Update and uninstall

Update litman

lit self-update upgrades litman through whichever tool installed it, uv or pipx. It prints current → latest and asks once.

litman also asks PyPI for the newest version number once a day, and prints a line when yours is older. LITMAN_NO_UPDATE_CHECK=1 switches that off — what it checks.

Uninstall litman

Two steps, in order, while the lit command still exists:

lit uninstall              # bundled skills, desktop shortcut, shell completion, vault registry, agent preferences, app-window browser profile
uv tool uninstall litman   # or: pipx uninstall litman

lit uninstall lists what it will delete and asks first — --dry-run previews, -y skips the prompt. Your vault (papers, PDFs, notes, annotations) is never touched by any of this; delete that directory by hand if you want the data gone.

Documentation

Full documentation lives under docs/. New to litman? The tutorial covers about 80% of everyday use.

Topic File
Start here — docs map docs/0-readme.md
Design philosophy docs/1-philosophy.md
Four-layer architecture docs/2-architecture.md
Concepts and field reference docs/3-concepts.md
Command reference docs/4-commands.md
Tutorial docs/5-tutorial.md
Agent model benchmark docs/6-agent-benchmark.md

Acknowledgments

Developed in the Süssmuth Lab, Technische Universität Berlin, with access to the TU Berlin HPC cluster.

Built with:

Claude Code Cursor

Standing on:

Click FastAPI Uvicorn ruamel.yaml pypdf Pydantic Rich httpx

Cloud sync runs on rclone, installed separately.

The litman wordmark is adapted from Nunito.

License

MIT. See LICENSE.


AI agents: a condensed, link-dense map of this project lives in README-Agent.md.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

litman-1.2.1.tar.gz (1.4 MB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

litman-1.2.1-py3-none-any.whl (1.5 MB view details)

Uploaded Python 3

File details

Details for the file litman-1.2.1.tar.gz.

File metadata

  • Download URL: litman-1.2.1.tar.gz
  • Upload date:
  • Size: 1.4 MB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.5

File hashes

Hashes for litman-1.2.1.tar.gz
Algorithm Hash digest
SHA256 4fb05274dd68c7653e611ba40a39bf503fc8735421f02ceeb5620fe456a08be2
MD5 b09eda09ad5fa50c2d91ea9b1c10480f
BLAKE2b-256 f291bfb5aef27740243202e39e18c341ed5f6d376a86f99a8140f5c5e4775fa8

See more details on using hashes here.

File details

Details for the file litman-1.2.1-py3-none-any.whl.

File metadata

  • Download URL: litman-1.2.1-py3-none-any.whl
  • Upload date:
  • Size: 1.5 MB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.5

File hashes

Hashes for litman-1.2.1-py3-none-any.whl
Algorithm Hash digest
SHA256 863dffdf2a183c0bb35e0f3fbc291efcff66c61adb245517eb4d52949826cd46
MD5 76057eee93c3078ff70773f2a5092ebf
BLAKE2b-256 37f53002220e0652491d1a60fd902a49fb1033b31277f86f8676cf0b27cda603

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page