Install and open
There is no install. Open the page and the workbench is running.
That sentence is the product, so this page spends its length on the part that actually needs explaining: where your work is kept, why the app asks for a folder, and what happens when you clear your browser data.
Opening it
Go to the site and press Open the workbench. That is the whole procedure. No account, no download, no extension, no upload.
What is happening underneath is worth one line, because it explains the rest of this page: every file you open is parsed by code running in your browser tab. There is no server doing the parsing, which is why there is nothing to upload, nothing to wait for, and no size limit imposed by someone else's request timeout.
It keeps working offline. Once the app has loaded, it is a normal installed-feeling web app: the service worker holds the application shell, so a lost connection stops nothing except the features that are explicitly remote — BLAST, the EBI services, and fetching a record from NCBI or UniProt. Those are network tools by definition; everything else is arithmetic on your own machine.
Do I need an account?
No, and signing in does not move your work. An account buys exactly three things:
- work you choose to publish gets an author,
- edits made by someone you shared a project with carry their name instead of yours,
- files you choose to upload get somewhere to live.
Nothing else changes. Projects, documents, files, layout, history, checkpoints and undo/redo are
device-tier state and never leave the machine — not on sign-in, not on share, not ever. The full
boundary is in AUTH_STORAGE.md.
Where your work is kept
This is the one piece of the model worth reading properly, because it is the opposite of what a web app usually does.
The folder on your disk is the source of truth. The browser's storage is a rebuildable cache.
On first run the lobby shows a strip offering to Choose a workspace folder. Take it. What you get is a real directory you can see in your file manager, back up, put in Dropbox, or open with anything else:
<your folder>/
index.melad the id → folder map
my-cloning-project/
project.melad the project: metadata + editor state, as a zip
pBR322.gb your real files, sitting there as themselves
insert.fastaTwo things follow from that layout, and both are the reason it exists:
- Your sequence files are still files. They are written beside the project, under their own names, in their own formats. Nothing is locked in a proprietary blob you need this app to read.
- "Clear site data" costs you one folder re-pick, never your work. The app rescans the directory and repopulates its caches. This is not a promise about our reliability — it is a property of where the bytes live.
If you never pick a folder
The app still works, and says what it is doing rather than pretending:
| Browser | What you get | Visible to you? |
|---|---|---|
| Chromium (Chrome, Edge, Brave, Arc…) | A real folder you choose, via the File System Access API | Yes — it is an ordinary directory |
| Firefox, Safari | OPFS — durable browser-owned storage | No, but it survives restarts |
| Anything older | In-memory, last resort | No, and it does not survive a reload |
The strip tells you which of these you are on. On Firefox and Safari your work is durable but private to the browser, so an export is your backup rather than a convenience — which is the honest reason the folder prompt exists on Chromium and the reason it cannot exist elsewhere.
Reconnecting
A browser hands back a folder permission that can lapse — after a restart, or when the browser decides to re-ask. When that happens the lobby shows Reconnect your workspace folder instead of the picker. It is one click and the same folder; nothing was lost in between, because the disk copy is the authoritative one.
On a phone or a small screen
The site reads and navigates at any width. The three words in the header — Tools, Formats and Docs — collapse into the menu button at the top right, and everything behind them is still there: all 53 tools grouped by bench task, the formats with what each one carries and whether it can be written as well as read, and the link into these guides. Tap a tool to read what it does before you commit a bench day to it.
Two honest caveats:
- The reading is complete; the editing is not designed for a phone. The workbench is a dense scientific surface — a canvas, a sequence, and a rail of panels side by side — and it assumes a screen wide enough to hold two of those at once. Use a small screen to browse what a tool does, to read a method, or to check a result someone shared with you. Do the work on a laptop.
- Storage is the browser's, not a folder. The workspace folder needs the File System Access API, which mobile browsers do not offer; on a phone your work goes to the browser's own durable storage, exactly as it does on desktop Firefox and Safari. See If you never pick a folder above — an export is your backup.
What a project is
A project is a folder, and inside the app it is files + a canvas + results + history kept together. You are not opening one sequence at a time: a project holds the plasmid, the primer list, the Sanger reads that came back, and the analysis you ran on them, and every tool acts on whatever is currently open.
That shared context is the design of the whole workbench, and it is what the workbench tour is about.
Next
- Your first project — a real sample from the lobby, end to end.
- Files and formats — what opens, and what each format keeps.
- The workbench tour — the canvas, the rails, the palette.
If you have a job to do rather than a tool to learn, skip all three and go to Workflows.