ProjectShelf for researchers and writers
Track papers, datasets, and long documents alongside the tools that open them, from Obsidian vaults and LaTeX projects to Zotero libraries and reference folders.
A paper is a project in every sense that matters here. It has a folder, a set of tools that open it, a status, and a habit of disappearing into a directory tree for months at a time. ProjectShelf is a plain inventory, so it does not care whether the folder holds source code or a literature review.
What to record
- Name
- Coastal erosion survey 2026
- Local path
- /home/you/research/coastal-erosion
- Repository URL
- The preprint, DOI, or shared drive
- Hosted URL
- The journal submission portal
- Tags
- paper, fieldwork, coauthored
- Status
- Active while writing, Paused while under review
- Notes
- Coauthors, deadlines, and where the raw data lives
The notes field is searchable, which makes it the right home for the details you will search by later: a collaborator name, a grant number, the conference you are targeting.
Opening writing tools
Most writing tools take a folder or file path exactly like an editor does. Add each one under Settings, then Applications, or as an editor if it is your usual destination for a project.
- Obsidian vault
- obsidian "obsidian://open?path={path}"
- VS Code with LaTeX
- code "{path}"
- Typora
- typora "{path}"
- Zotero
- zotero
- RStudio project
- rstudio "{path}"
- Jupyter Lab
- cd "{path}" && jupyter lab
A LaTeX build command
Because commands are templates, you can store the build alongside the project rather than remembering it. Add this under Settings, then Commands, in the build category.
cd "{path}"
latexmk -pdf -interaction=nonstopmode main.texSelect several papers and this generates the build for each of them as one block, which is handy before a deadline when you want every draft regenerated.
Datasets and large files
Record the dataset as its own project with its own path and a tag such as dataset. Then the papers that depend on it can name it in their notes, and searching the dataset name surfaces everything connected to it. ProjectShelf stores no file contents, only the location, so a multi-gigabyte dataset costs nothing to track.
Launch sets per working session
Research sessions tend to need several things at once: the manuscript, the analysis folder, the reference library. Put those in a launch set named for the session and launching it generates the commands to open all of them together.
Backups matter more here
Everything lives in your browser, so clearing site data would take the inventory with it. Export a JSON backup from Settings, then Data, and keep it alongside your other research files. It restores on any machine, which also makes it a reasonable way to move between a laptop and a lab workstation.
ProjectShelf runs entirely in your browser. There is no account and nothing to install before you try it.
Open ProjectShelf