GridPluck is a free, open-source Chrome extension that lets you hover over structured data, select it, and reuse the currently loaded rows without fighting broken copy-and-paste.
- For spreadsheets: copy TSV or download an Excel-friendly CSV.
- For AI agents: copy a Markdown table or download structured JSON.
- For your privacy: no account, analytics, backend, uploads, or extraction history.
- Click the GridPluck toolbar icon.
- Hover until the data you want is outlined.
- Click to preview the extracted rows.
- Copy TSV or Markdown, or download CSV or JSON.
GridPluck reads only rows already loaded in the page. It does not click through pagination, trigger infinite scrolling, or crawl in the background.
GridPluck is not yet distributed through the Chrome Web Store.
- Download
gridpluck-v1.0.0.zipfrom the latest GitHub release. - Unzip it.
- Open
chrome://extensions. - Turn on Developer mode.
- Choose Load unpacked and select the unzipped folder.
- Pin GridPluck from Chrome's Extensions menu.
Chrome displays GridPluck's two permissions: activeTab and scripting. They let it run only after you click the icon in the current tab. GridPluck does not request persistent access to every website.
| Page structure | Support |
|---|---|
Semantic HTML <table> |
Yes, including rowspan and colspan |
| ARIA table/grid/treegrid | Yes |
| Repeated div-based rows | Heuristic detection |
| Flat CSS Grid | Heuristic detection |
| Rows not loaded in the DOM | No, by design |
| Closed shadow roots | No |
| Browser-protected pages | No |
Visual layouts are messy. GridPluck rejects low-confidence structures instead of pretending an entire page is a useful table. If a public page fails, open a bug with the URL or contribute a minimal, sanitized fixture.
Requirements: Node.js 20+ and Google Chrome.
git clone https://github.com/devashishPM/GridPluck.git
cd GridPluck
npm ci
npm run check
npm run packageLoad the generated dist/ directory from chrome://extensions. During development, npm run dev rebuilds when source files change; reload the extension to apply an update.
- Visible means user-controlled. GridPluck extracts the table you point to, not whatever a crawler guesses you wanted.
- Faithful before clever. The preview is read-only. GridPluck does not silently clean, transform, or enrich data.
- Local is a product feature. Export logic runs in the tab and the extension contains no network client.
- Small permission surface. No
<all_urls>, storage, clipboard, identity, or network permissions. - Test the weird parts. Merged cells, ragged rows, accessible grids, hostile strings, and page-style isolation are fixtures—not promises in prose.
Read the architecture, roadmap, privacy statement, and contribution guide.
Bug reports and focused pull requests are welcome. Table extraction has an unusually long tail, so the most valuable contribution is often a small fixture that reproduces a real structure without including private data.
Start with the contribution guide and choose an issue labeled good first issue.
GridPluck was created by Devashish Shrivastava, a product leader who kept encountering useful tables across product surfaces with no export option—whether the next step was a spreadsheet or structured context for an AI agent.
GridPluck is available under the MIT License. Required upstream license notices are included in THIRD_PARTY_NOTICES.md.