Changelog

Real product changes with their real dates — this list is curated from the project's commit history. Entries are never backdated or re-dated.

2026-08-16 — Verified video and audio conversion family

• Added WebM to MP4, MOV to MP4, MP4 to MP3, WebM to MP3, M4A to MP3 and WAV to MP3 as real native FFmpeg conversions.
• Published fixed profile 1.0.0: H.264/yuv420p + AAC for MP4 and 192 kbps stereo for MP3; no user-controlled FFmpeg arguments.
• Added signature checks before conversion, ffprobe verification after conversion, a complete output decode, and codec/duration-tolerance gates.
• Published five downloadable synthetic inputs, six expected outputs and a six-case JSON/CSV validation dataset; per-job reports are exportable from each download page.
• Added explicit 100 MB, 10 minute, rate, concurrency and temporary-file cleanup disclosures to every media page and the security documentation.

2026-07-29 — Evidence-first File Intelligence suite

• Added eight interactive tools: File Health Check, Metadata & Privacy Inspector, PDF Accessibility Checker, Scanned PDF to Searchable PDF, Sensitive Data Finder, Invoice & Receipt extraction, Semantic Document Compare and Document Assistant.
• Every tool publishes a real sample file and example report, its calculation method, supported and unsupported cases, confidence meaning, comparison table, known false-result conditions and a versioned accuracy log.
• Reports can be downloaded as JSON or CSV. Sensitive-data exports mask candidate values and never include the complete extracted document text.
• Published an anonymous 41-case synthetic release dataset with raw CSV; the release test fails when the public case counts and actual assertions diverge.
• Optional AI is explicitly secondary: the deterministic local report runs first, and only extracted text is sent after a separate user action.

2026-07-22 — A file platform, not just a converter

• File Health Report: dropping a file now shows what it really is — content-based type detection, extension-vs-content mismatch warnings, image dimensions/transparency/animation, PDF version/encryption/truncation, CSV dialect and JSON validity — all read from the file header in your browser.
• Smart Settings Advisor: pick a goal (email, web, print, smallest, archive) and get a recommended setting with its reason and trade-off; nothing changes until you press Apply.
• Quality Scorecard on every result: output size, size change and time up front, honest notes when a file gets bigger, plus JSON/CSV export and a print view.
• Before/after comparison: a split-slider with SSIM/PSNR similarity for images, and a page viewer that marks selectable-text loss for PDFs.
• Structured error help: failures now end with a plain-language cause, one next step and a link to the tool that fixes it.
• Conversion recipes and saved presets: chain steps into one drop, and save your settings per tool (export/import as JSON, no account).
• Per-tool "preserves / changes / limits" facts on all 80 tools, each checked against the engine code.
• Conversion Labs and Guides: published measurements from our own engines with downloadable raw data, and task-first how-to guides.

2026-07-22 — Ad policy overhaul and build-time quality gates

• Download/result screens, legal pages and error pages no longer carry any advertising — zero ad requests on those screens.
• Removed the sticky sidebar ad and the mobile anchor ad site-wide; tool pages are capped at two in-content ad units — the first only after the how-to and why sections, the second at the end of the editorial block, nothing next to the upload area.
• The build now fails on policy violations: missing titles or canonicals, duplicate titles, ads on ad-free screens, broken internal links and near-duplicate pages.
• Each category page received its own editorial content and FAQ; the About page's unverifiable audience claim was removed and an honest how-the-tools-are-built section was added.
• Added this trust section: processing transparency per tool, editorial policy, testing methodology, changelog and status.
• Hardened the product to match its claims: Protect/Unlock verify their output on every run, Redact re-extracts the output text layer and fails if anything survived, server-conversion file names moved out of request URLs, and pre-consent pageviews are counted anonymously.

2026-07-11 — Editorial depth on every tool page

• Hand-written editorial sections on all 80 tool pages.
• Download pages explain practical next steps for each output format.
• Server hardening: malformed URLs no longer crash the request handler.

2026-07-10 — Brand and URL hygiene

• The visible brand, canonical URLs, legal copy and sitemap now derive from the selected deployment target.
• Extensionless page URLs 301-redirect to their canonical trailing-slash form.
• Conversion progress is shown with a real progress indicator; the file queue collapses into a compact "add more" bar once files are loaded.

2026-07-09 — The big capability round: 51 → 80 tools

• PDF organizing suite: merge, split, rotate, remove, extract, reorder, page numbers, watermark, crop.
• Image editing tools: compress, resize, crop, rotate/flip, watermark.
• Compress PDF, Sign PDF, Compare PDF and OCR (images and scanned PDFs).
• Server-side Word/Excel/PowerPoint/HTML to PDF via LibreOffice and headless Chromium.
• PDF to Word, PDF to PDF/A, Repair PDF, Protect PDF (AES-256), Unlock PDF and true Redact PDF (removed content is deleted, not covered).
• Twelve engine defects found by an internal quality audit were fixed and re-verified.

2026-07-04 — Universal image hubs

• Image to WebP/GIF/BMP/ICO hubs accept any supported picture format in one drop.
• Related-converters block moved directly under each tool.

2026-07-02 — Launch

• 47 browser-side converters for images, PDFs and data formats, with batch conversion and ZIP download.