ToolForge
fully responsive Document & File Toolkit that runs entirely in the browser and can be deployed on GitHub Pages. The application must not require any backend or server. Every operation should happen locally on the user's device for privacy.
- Language
- TypeScript
- Stars
- 0
- Last updated
- License
- MIT
From the README
93 document, image, text and utility tools that run entirely in your browser. No uploads, no accounts, no tracking — and it keeps working with no connection at all.
ToolForge is a static single-page app. Once the page has loaded there is no server left to talk to: every merge, render, hash and conversion happens inside your browser tab using Canvas, Web Crypto, IndexedDB and a handful of well-known client-side libraries.
Live demo · MIT licensed
Screenshots
| Home | Tool workspace |
|---|---|
![]() |
![]() |
| Page organiser | Light mode |
|---|---|
![]() |
![]() |
Screenshots are placeholders — drop your own PNGs into
docs/screenshots/after the first run.
Features
Categories overlap slightly — Compress PDF is listed under both PDF and Compression, for example — so the lists below add up to a little more than the 93 distinct tools.
Merge · Split · Extract pages · Delete pages · Rearrange (drag & drop, undo/redo) · Rotate · Crop · Duplicate pages · Insert blank pages · Compress · Add watermark · Remove watermark · Add page numbers · Metadata editor · Flatten · Sign (draw or upload) · Draw signature · Add text · Highlight · Underline · Sticky notes · Fill AcroForms · Password protect · Unlock · View · Search text · Extract text · Extract embedded images · PDF to images · Images to PDF · Thumbnail / contact-sheet generator · OCR (opt-in, in progress)
Images
Resize · Compress (by quality or target size) · Crop (drag or numeric) · Rotate · Flip · Convert to JPG / PNG / WebP / SVG · GIF conversion · Image to PDF · Watermark · Blur · Grayscale · Brightness · Contrast · Saturation · Background colour · QR generator · QR scanner (file or camera)
Text
Word counter · Character counter · Remove duplicate lines · Remove extra spaces · Sort lines · Reverse text · Base64 encode/decode · URL encode/decode · JSON formatter · JSON validator · XML formatter · HTML formatter · Markdown preview
Conversion
JPG/PNG/WebP to PDF · PDF to PNG/JPG · ZIP creator · ZIP extractor · CSV to JSON · JSON to CSV · Markdown to HTML
Security
Password generator (with entropy readout) · SHA-256 · SHA-512 · MD5 · UUID v4 · JWT decoder · JWT viewer with local HS256 verification · Random string generator · PDF encryption and decryption
Utilities
QR code generator (text, URL, Wi-Fi, vCard) · Barcode generator (CODE128, EAN, UPC and more) · Colour picker with contrast checks · Colour palette generator · Timestamp converter · Unix timestamp clock · Lorem ipsum · Random numbers · Dice roller · Coin flip
Experience
- Responsive from 320 px to ultrawide, with a mobile drawer and a desktop sidebar
- Dark, light and system themes, plus six runtime accent colours
- Command palette (Ctrl/Cmd + K) with ranked search
- Drag & drop, multi-file selection and paste-to-upload everywhere
- Favourites, recently used tools and a local download history
- Toasts, progress bars, loading skeletons and smooth page transitions
- Batch processing with Download all as ZIP
- Drag-to-reorder pages and files; undo/redo in the page organiser
- File-size analytics and estimated compression savings on every result
- Installable PWA — works fully offline after the first visit
- Five UI languages (English, Spanish, French, German, Arabic with RTL)
- Keyboard navigable, focus-visible rings, skip-to-content link, ARIA labels
Privacy
The privacy claim is structural, not a promise: there is no backend to send anything to. The only network requests the app makes are for its own static assets, all of which are precached by the service worker on first load. You can verify this in DevTools -> Network, or by turning off Wi-Fi and using the app normally.
Local storage holds preferences, favourites and download history (file names and sizes, never contents). IndexedDB optionally caches recently opened files so you can re-run a tool without picking the file again; Settings -> Clear all local data removes everything.
Installation
Requires Node 20+.
git clone https://github.com/abdulmanan69/pdf-toolkit.git
cd pdf-toolkit
npm installLocal development
npm run dev # start Vite on http://localhost:5173 npm test # run the unit test suite npm run typecheck # strict TypeScript check npm run build # production build into dist/ npm run preview # serve the production build locally npm run icons # regenerate the PWA PNG icons
predev and prebuild copy PDF.js's character maps and standard fonts into public/pdfjs/
so CJK documents render correctly offline. That folder is generated, and git-ignored.
GitHub Pages deployment
The repository ships with .github/workflows/deploy.yml, which type-checks, tests, builds and
publishes on every push to main.
- Push the repository to GitHub.
- Go to Settings -> Pages and set Source to GitHub Actions.
- Push to
main. The workflow builds and deploys automatically.
How the routing and SEO fit together
The site serves real URLs — /pdf-toolkit/tool/merge-pdf, not #/tool/merge-pdf — because
hash fragments are invisible to crawlers. GitHub Pages has no rewrite rules, so the build does
the work instead:
prebuildregeneratespublic/sitemap.xmlandpublic/robots.txtfromsrc/tools/catalog.json, so the sitemap can never drift from the tool list.postbuildprerenders one realindex.htmlper route (102 of them) intodist/, each with its own<title>, meta description, canonical URL, Open Graph tags and JSON-LD (SoftwareApplication+BreadcrumbList).- A
dist/404.htmlcopy catches anything unmatched so the router still boots.
The result: every tool URL is a genuine 200 with unique metadata before a single line of
JavaScript runs, and useSeo in src/seo.ts keeps those same tags correct during client-side
navigation.
If you rename the repository, change base in vite.config.ts and SITE.url in src/seo.ts
and SITE in scripts/seo.mjs — those three constants drive everything else.
Technologies used
| Area | Choice | Why |
|---|---|---|
| Framework | React 18 + TypeScript | Component model, strict typing |
| Build | Vite 6 | Fast dev server, easy static output |
| Styling | Tailwind CSS 3 | Token-driven design system via CSS variables |
| Routing | React Router 6 (HashRouter) |
Static-host friendly |
| Animation | Framer Motion | Layout transitions and shared-element pills |
| PDF writing | @cantoo/pdf-lib |
A maintained pdf-lib fork with the identical API plus AES encryption/decryption, which upstream pdf-lib does not support. Password Protect and Unlock need it, and using a single copy keeps the bundle small. All PDF writing goes through src/lib/pdf.ts, so swapping back is a one-file change. |
| PDF reading | PDF.js | Rendering, text layer, embedded image extraction |
| Archives | JSZip | ZIP create/extract and batch downloads |
| Downloads | FileSaver.js | Cross-browser save dialogs |
| Codes | qrcode, jsqr, jsbarcode |
Generation and offline scanning |
| Markdown | marked |
Preview and HTML export (output is sanitised) |
| Icons | lucide-react |
Consistent icon set |
| Tests | Vitest + jsdom | Unit tests for the pure utility layer |
Folder structure
.
├── .github/workflows/deploy.yml # Pages CI/CD
├── public/ # Static assets (icons, robots, sitemap, PWA)
├── scripts/
│ ├── generate-icons.mjs # Dependency-free PNG icon generator
│ └── copy-pdfjs-assets.mjs # cmaps + standard fonts -> public/pdfjs
└── src/
├── components/
│ ├── Layout.tsx # Shell, sidebar, command palette, error boundary
│ ├── PdfPageGrid.tsx # Thumbnail grid: select, reorder, rotate
│ ├── ui.tsx # Design-system primitives
│ └── workbench.tsx # FileWorkbench / TextWorkbench / Dropzone
├── context/AppContext.tsx # Preferences, toasts, favourites, history
├── lib/
│ ├── crypto.ts # SHA family, MD5, UUID, JWT, password entropy
│ ├── image.ts # Canvas pipeline (crop/resize/filter/watermark)
│ ├── pdf.ts # pdf-lib helpers (create, merge, embed, save)
│ ├── pdfjs.ts # PDF.js helpers (render, text, images)
│ ├── storage.ts # localStorage + IndexedDB
│ ├── text.ts # Pure text/CSV/JSON transforms
│ └── utils.ts # Formatting, downloads, page-range parsing
├── tools/
│ ├── registry.ts # Catalogue: metadata + lazy components
│ ├── pdf-organize.tsx # Merge, split, extract, rotate, crop
│ ├── pdf-content.tsx # Watermarks, numbering, metadata, signing, forms
│ ├── pdf-annotate.tsx # Text, highlight, underline, sticky notes
│ ├── pdf-view.tsx # Viewer, search, extraction, rendering
│ ├── pdf-secure.tsx # Encrypt, decrypt, compress
│ └── image.tsx text.tsx convert.tsx security.tsx utility.tsx
├── hooks.ts # Reusable hooks (hotkeys, undo/redo, clipboard)
├── i18n.ts # UI translations + RTL handling
├── pages.tsx # Routes: home, category, tool, settings
└── types.ts # Shared types
How a tool is built
Adding a tool is usually about 30 lines. FileWorkbench owns the whole file loop — drag & drop,
batching, progress, errors, per-file download and ZIP download — so a tool only describes its
options and an async run:
export function GrayscaleImage() { const [amount, setAmount] = useState(100); return ( <ImageTool toolName="Grayscale" actionLabel="Apply grayscale" ops={{ filters: { grayscale: amount } }} optionsKey={amount} options={<Slider label="Grayscale" value={amount} onChange={setAmount} min={0} max={100} />} /> ); }
Then register it in src/tools/registry.ts with an id, icon, category and lazy import.
Known limits
Being honest about a browser-only design:
- Very large files are bounded by device memory, not by an upload quota. Rasterising a 500-page scan is slower here than on a server.
- Remove Watermark covers matching watermark text and deletes watermark/stamp annotations. Marks burned into a scanned image cannot be removed.
- Convert to SVG embeds the raster in a scalable SVG container; it is not vector tracing.
- Convert GIF exports the first frame — animation is not preserved.
- Unlock PDF decrypts a document you already have the password for. It does not crack passwords.
- OCR is marked Coming soon: a usable offline model is 10–15 MB per language, so it will ship as an opt-in download rather than bloating first load.
Contributing
- Fork and create a branch:
git checkout -b feat/my-tool. - Keep
npm run typecheckandnpm testgreen. - Add unit tests for anything in
src/lib/— that layer is pure and fully testable. - Match the existing conventions: tools live in a category module, get an entry in
registry.ts, and reuseFileWorkbench/TextWorkbenchinstead of re-implementing uploads. - Never add a network call. A pull request that sends user data anywhere will be declined.
- Open a PR describing the tool, the libraries it needs and its bundle impact.
Author
Abdul Manan
- GitHub: @abdulmanan69
- Email: manan999999999@gmail.com
Designed and built the whole toolkit — all 93 tools, the design system and the offline-first architecture.
License
MIT © Abdul Manan — see LICENSE.



