Turn a company directory into a call list
Two hundred companies, spread over ten pages, and the sheet is due today.
List ExtractorChrome extension, MCP server, cloud runs
Extract data from any website. Click a list, get a spreadsheet, or ask Claude.
Tabitha reads the page you already have open. Pick one item in a list and it finds the rest, turns the pages and hands you a table. That works without an API key, and without a script you have to repair every time the site changes its markup.
The extension is free. It runs in your browser, on your own session.
What people scrape with Tabitha
Eight of the jobs people bring it on the first day. The click is the same on all of them and only the page changes.
Copy a competitor catalogue into a sheet
Their shop shows twenty products per page. You want the whole assortment next to yours, in one view.
List ExtractorWatch competitor prices without a subscription
You need to know when they cut a price. Tools that do this start at a monthly fee per shop.
List ExtractorRead every review at once
Four hundred reviews sit behind a load more button. What you want is the thing people repeat, and the first ten will not tell you.
List ExtractorCollect job ads from a board
You are tracking who hires for what. The board shows ten offers at a time and has no export button.
List ExtractorCompare repositories and dev tools in a sheet
You are choosing a library and the shortlist lives in twenty browser tabs.
List ExtractorPull a Shopify catalogue ready for import
You are moving a shop, or measuring one against yours. Either way the import file wants variants, prices and stock in a particular shape.
Shopify ExtractorHand the file over in the shape the next system wants
Your CRM rejects the import because the price column is text and the same company appears three times.
Data TableOne scraper, three ways to run it
The extraction logic is the same in all three. They differ in who presses the button and where the page loads.
In the extension
Open the side panel, click one row on the page, check the columns Tabitha guessed, run it. Results stay in your browser until you decide to sync them.
See the nine toolsAsk Claude
Connect Tabitha over MCP and describe the job in words. Claude starts the run in the browser you have open, then reads the result back as rows and counts. The page source stays in the browser.
Use it with ClaudeIn the cloud
Hand a saved recipe to our server for a nightly schedule or a long list of URLs. The table lands in the same account you use in Chrome.
How cloud runs workThree of the nine tools
Each one does a single job and they write into the same table, so you can start with a list and finish with contact details.
All nine toolsWhy sites usually let it through
Nobody can promise you will never be blocked, and a scraper that promises it is selling you a story. What you get instead are five habits, and you can judge them yourself.
No synthetic input events
Most extensions fake mouse moves and clicks. A faked event arrives with isTrusted set to false, and one line of JavaScript on the page can read that flag. Tabitha does not dispatch them. It navigates, scrolls and focuses the way the browser itself does.
A human rhythm
Delays come from a log normal distribution, with a persona drawn once per session. Reading time grows with the amount of text on the page. Longer pauses turn up now and then, the way they do when a person gets distracted.
A budget for every host
One tab per site, a cap on requests per hour, backoff after a 429 or a 503, and crawl delay from robots.txt treated as the floor rather than a suggestion.
It stops at a verification page
When a challenge, a captcha or a 403 shows up, the run pauses and asks you to pass it yourself in the same tab. Then it picks up where it left off, without trying to outrun the check.
Your session, your cookies
Pages behind a login work because the page is loaded by your browser, with your account, not by a rented profile in a datacenter.
Trusted input through the Chrome debugger exists as an opt in for pagination clicks that refuse to work any other way. It is off unless you turn it on, and Chrome shows a banner while it runs.
Give Claude a scraper
Tabitha runs an MCP server over Streamable HTTP. Add it once and Claude can list your tables, filter them, deduplicate them, export them and start a run in the browser you have open.
Claude Code, one command
claude mcp add --transport http tabitha https://app.tabitha.creathink.one/mcp \
--header "Authorization: Bearer <token>"The token comes from your account, under Settings and Tokens. Claude Desktop and Cursor take the same URL in their own config file.
Ten tools the agent gets
- list_tables
- get_table_schema
- query_table
- clean_table
- export_table
- list_recipes
- run_in_browser
- run_in_cloud
- get_run
- cancel_run
Cloud runs, with the fine print up front
Cloud runs are not switched on yet. When they are, you save a recipe, hand it to the server and get a schedule, a webhook and a run history. The two limits below decide whether that fits your job.
- The cloud runs from one server in one datacenter in Germany. There is no proxy pool, no residential addresses and no country picker.
- Sites that reject datacenter traffic will reject it there too. For those, run the same recipe in your browser, where the request comes from your own connection.
Questions people ask first
Do I need to know how to code?
No. You click an item in a list and confirm the columns. Tabitha works out the selectors, the pagination and the column types, and you can rename or drop a column before the run starts.
Where is my data kept?
In your browser, in IndexedDB, until you sign in and sync a table on purpose. Synced tables sit in a PostgreSQL database on our own server in Germany. Nothing is sent to a third party analytics or enrichment service.
Is scraping legal?
It depends on what you collect, whose terms you accepted and what you do with the result. Public pages are not a free pass. If the data describes people, GDPR applies and you are the controller, which means you need a lawful basis and you have to answer to the people in your table. Tabitha does not make that call for you. Read the terms before you build a lead list.
Does it work on pages behind a login?
Yes, when you run it in the extension. The page loads in your tab with your session, so whatever you can see, Tabitha can read. In the cloud you get only what is public.
What happens when Cloudflare shows up?
The run pauses and tells you which tab needs your attention. You pass the check by hand, then the run continues from the page where it stopped, without a faked click or a retry loop.
Which websites are supported?
There is no per site list, because there are no per site adapters. If a page shows repeating items in HTML, the list tool works on it. Shopify stores get a dedicated tool that reads the store product feed instead of the page.
Can I run something every night?
Not yet. Cloud runs are the piece that does it, and they are still being built, so nothing on that is on sale. The extension runs while your browser is open.
Do I need an account?
Not for extracting and exporting in Chrome. You need one, on the Pro plan, to sync tables between devices and to use MCP.
What does it cost?
The extension is free and stays free: nine tools, nothing metered, export included. Pro costs 54 USD a year net and adds the account, table sync and MCP. Cloud runs are not for sale yet, because they are not built yet.
Is it in the Chrome Web Store?
Not yet. For now you install the zip in developer mode, which takes about a minute. The download page walks through it.
Take the first table today
Install the extension, open a list you already work with and see how many columns come back on the first try.
Download for ChromeThe extension is free. Pro costs 54 USD a year.