WebToolsUltima

Every tool you need, running in your browser. A local-first, server-free tool platform: a growing library of fast, private utilities that run entirely on your device. Convert & compress files, encode & hash text, generate data, crunch numbers — with no uploads, no accounts, and no servers. Your data never leaves your machine.

Powered by or derived from software by Cognitive Industries.

See CLAUDE.md for the standing rules and TOOL_LOG.md for the engine/tool registry.

Why it's different

WebToolsUltima isn't a wall of one-off pages backed by a server farm — it's a single client-side tool engine that everything plugs into:

• An any → any file engine. Formats are graph nodes; converters are edges. A planner runs a shortest-path search to get from your input to whatever you want, chaining steps when there's no direct route (e.g. HEIC → PNG → JPG happens automatically). Tools like "WEBP → PNG", "MP4 → MP3" or "PDF Merge" are thin presets over the graph — add a capability to an engine and new tools appear for free.
• Instant utilities, too. Beyond files, WebToolsUltima hosts standalone widgets — Base64 / URL encoders, JSON formatter, SHA hashes, case converter, UUID / password / lorem generators, unit & time converters — all running locally.
• Everything is private and off the main thread. Heavy work runs in Web Workers; large engines (ffmpeg.wasm) lazy-load only when a tool needs them. No file bytes are ever sent anywhere — verify it yourself in the Network tab.

Tool families

Group	Categories	Examples
Files	Convert · Compress · Modify	Image/Audio/Video/PDF/Doc convert, image & PDF & media compress, crop/rotate/resize, PDF merge/split, GIF maker, video trim
Utilities	Text & Code · Generate · Calculate	Base64, URL encode, JSON formatter, hashes, case convert, counters · UUID / password / lorem · unit & time converters

Architecture

src/
  core/
    formats/      Format registry + content-sniffing detector
    engines/      Engine/Capability contracts + engine registry (capability index)
    planner/      Graph search → a Pipeline of steps
    pipeline/     Executor (batch, fan-in/out, progress, cancel) + worker pool
    jobs/         Job/activity store (live progress, cancel)
    util/         bytes, blob, zip, download helpers
  engines/        Self-contained capability providers (canvas-image, pdf, archive, heic,
                  vector, gif, ffmpeg, docs, data). Each lazy-loads its own deps.
  tools/          Tool presets derived from capabilities + curated named tools; categories
  ui/
    components/preview/   FilePreview + interactive editors (crop/trim) bound to params
    widgets/              Instant non-file tools (encoders, hashes, generators, calculators)
    ...                   Launcher (search grid) + tool view (dropzone → options → run → results)
  workers/        Shared transform worker hosting the light engines

Two extension points

// 1. A file Engine — adds capabilities (edges/operations) to the graph
interface Engine {
  id: string;
  capabilities: Capability[];      // convert | compress | transform
  load(): Promise<EngineRuntime>;  // lazy: import libs / init wasm
}

// 2. A widget — a standalone client-side tool, registered in ui/widgets/registry.tsx

Adding a file tool = adding a capability (or engine) + a preset entry; the planner and UI pick it up automatically. Adding an instant tool = a component + a registry entry. A startup assertion guarantees the catalog never advertises something the engines can't actually do.

Develop

npm install
npm run dev         # http://localhost:5173 (sets COOP/COEP headers for ffmpeg threads)
npm run typecheck   # strict type-check
npm run build       # production build → dist/
npm run test        # vitest unit tests (planner, detection, engines, catalog)

Hosting note. The threaded ffmpeg.wasm build needs cross-origin isolation. Serve the production build with Cross-Origin-Opener-Policy: same-origin and Cross-Origin-Embedder-Policy: require-corp headers (the dev/preview servers set these automatically via a Vite plugin).

Privacy

WebToolsUltima makes no network requests with your data. The only fetches are the app's own static assets and lazy engine payloads (e.g. ffmpeg.wasm). You can confirm this in your browser's DevTools Network tab during any operation.

Roadmap

• Higher-fidelity document conversions (PDF↔Word reflow, rich EPUB/DOCX layout).
• Password-protect / unlock PDF (needs a PDF crypto layer, e.g. qpdf-wasm).
• More image codecs via WASM (mozjpeg, oxipng, AVIF encode).
• More instant tools (QR codes, colour tools, regex tester, diff, cron, …).
• Saveable pipeline recipes + PWA offline install.

License & attribution

Licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) — see LICENSE. You may use, study, modify and share this software non-commercially, provided you give credit and license derivatives under the same terms; commercial use requires separate written permission. Any use, fork, deployment or public availability must keep the attribution:

Powered by or derived from software by Cognitive Industries (https://cognitive-industries.org)