Utilities

Sometimes you do not need a new design. You need a file fixed: a photo in the wrong format, a PDF that carries more than it should, or two versions of a spreadsheet to compare. The Utilities tab (#/u) collects these jobs. They work on your device, and the file you bring is not uploaded.

signed by Lolly2026-10-21vector SVG124 paths~29k nodes324 groups395 KBTự kiểm chứngTải tệp đã ký signed by Lolly2026-10-21vector SVG124 paths~29k nodes324 groups396 KBTự kiểm chứngTải tệp đã ký

Find the right utility

• Share a file safely: Prepare for sharing, Strip Hidden Data and Redact.
• See what changed between two versions: Compare.
• Change a format or a size: Convert, Compress, Convert Image and Convert Font.
• Get the words or pictures out: Unpack, Text or Read and extract.
• Check where a file came from: Verify & Inspect.
• Rearrange or sign PDF pages: Pages and Sign.
• Cut or clean audio and video, or record your voice: Trim, Clean, Voice Recorder and Script audio.
• Check a colour or build a palette: Colour Lab, Contrast and Palette Lab.
• Read or fix a spreadsheet: Spreadsheet.
• Move a deck onto your design system: Rebrand and Theme swap.
• Everything else: screenshots, QR scanning, a countdown, a one-link page and more, under More utilities.

Built into the app, or a tool

Most utilities are tools. A tool keeps its inputs in the URL, so it saves sessions, reopens from a link, renders from the CLI and exports through the shared pipeline.

The rest are views: pages of the app with no manifest and no export panel. Compare, Prepare for sharing, Verify & Inspect, Unpack, Read and extract, Rebrand, Colour Lab, Spreadsheet and Script audio each have a card. Convert has none; open it from Search by typing "convert", or go to #/convert.

A view keeps no session, but it has room: it can use the app's own controls, run as a tall page and hold a 200,000-row grid or a 400-page document. It does not keep the file you drop. You take away a download or a copy: Convert keeps finished copies in Recent file operations, Prepare for sharing can Add to library and Rebrand keeps each deck as a project.

Every tile can be starred (Your favourites) and has a details dialog. A view's tile has no keep offline or saved sessions, because views ship inside the app.

Share a file safely

Prepare for sharing

Prepare for sharing (#/prepare) looks through text, JSON, YAML, HAR and ZIP files on your device and suggests what to replace before you share them. You choose the replacements, review each original beside its copy, then copy, download, send or save the result to the library.

Suggestions cover credential fields and common token shapes, email, payment-card, bank-account, phone, date, postcode, address and possible name patterns. They can be wrong, and they can miss confidential information. Keep all and Continue unchanged stay available. Add a missed exact value or a sensitive field name, edit a replacement or choose individual occurrences. Identical values share one replacement across every file in the job.

You can also start from elsewhere. Text opens the same panel for the selected input text, or all of it when nothing is selected. Files in Assets and saved file results have a Prepare for sharing action, and exported file utilities offer it for the copy.

lolly prepare request.har
lolly prepare request.har --review-file=private-review.json
lolly prepare request.har --choices=private-review.json --output=prepared.har
lolly prepare one.json two.yaml --replace-all --out-dir=prepared
lolly prepare image.png --strip-hidden-data --output=prepared.png

Strip Hidden Data and Redact

Strip Hidden Data removes selected DOCX properties, or image and PDF metadata, and lets you inspect other Office hidden content. Redact blacks out sensitive content by rebuilding the file, then checks the output before it downloads. Both are tools, so they save sessions like any other.

signed by Lolly2026-10-21vector SVG66 paths~9.4k nodes47 groups143 KBTự kiểm chứngTải tệp đã ký signed by Lolly2026-10-21vector SVG66 paths~9.4k nodes47 groups143 KBTự kiểm chứngTải tệp đã ký

See what changed

Compare (#/compare) puts two versions side by side on your device. It compares text, JSON, YAML, tables (CSV, TSV or XLSX) and image or PDF-page previews. For DOCX, PPTX and PDF files it compares the extracted text. The sources stay unchanged. Compare does not merge, apply edits, save the inputs or send their contents anywhere.

The same comparison opens from Assets and from file-operation review, including a saved copy with a reselected original whose hash must match. Compare table files in Spreadsheet and Compare document text in Unpack open it too. Changed values start folded, and a downloaded report holds only the summary until you select Include source names, paths and values. The inputs never enter the address bar or a default report.

Change a format or a size

Convert

Drop or choose a file, or a batch of up to 20 images, pick an output format and make a copy. Every conversion runs in the app, with no upload and no service. Finished copies stay in Recent file operations, including after a reload, ready to download or share. Your originals are not kept.

signed by Lolly2026-10-23vector SVG66 paths~11k nodes46 groups162 KBTự kiểm chứngTải tệp đã ký signed by Lolly2026-10-23vector SVG66 paths~11k nodes46 groups162 KBTự kiểm chứngTải tệp đã ký

For images, Quick presets, Quality, Longest edge (px) and Target file size (KB) shape the copy. The file chooser lists images, fonts, video, documents and data, up to 128 MB per file. A .txt or .md file is not in its list, so drop it onto the view instead. Batches of several files work for images only.

Compress, Convert Image and Convert Font

These three are tools, so they save sessions and travel in a link.

• Compress shrinks a PDF by recompressing its images.
• Convert Image is the photo path. It decodes HEIC and HEIF with a bundled decoder and writes WebP, JPEG or PNG, with a quality setting and a longest-edge resize. The Convert view has image settings of its own and also handles fonts, SVG, documents, video and data, which the tool does not.
• Convert Font swaps a font between TrueType, OpenType and WOFF.

Get the contents out

Unpack

Most software treats a design file as one sealed thing. Unpack (#/unpack) opens the container: the words, the vector marks, the images, the fonts, the colours and the attachments, each one viewable and downloadable. Nothing is uploaded, and the older #/pdf link still works.

Drop a PDF or Illustrator file, an SVG, an HTML page, an InDesign .idml, a Penpot .penpot, a Figma .fig, a PowerPoint .pptx or a Photoshop or GIMP .psd, .psb or .xcf. What comes out depends on what the format holds:

A names-only font row cannot be downloaded, because the file holds no font bytes.

The Text pass rebuilds each page in reading order with the page's own vector picture next to the text. It notes how many columns it read and how many rotated runs it left out, so you can judge the result. Take the words away with Copy all, Download .md or Download .txt, or one page at a time. An SVG whose type was outlined to paths yields no text, the same answer a PDF gives when its text was converted to curves.

Scanned pages have no text to extract. A born-digital PDF contains its glyphs and their positions, which is why extraction works offline. A scanned page holds a picture of text, and Unpack reports it as that, page by page, with a banner when every page is a scan. For those, use the on-device text reader: drop the image into Verify and press Read the text in this image, or open it in Assets and press Read text.

The tabs sit in a strip pinned to the top of the report, and each appears only when a pass found something:

• Palette: the distinct colours the file paints with, each with a copy button.
• Logos: vector marks, downloadable as SVG and sharp at any size. In a guidelines file these are often the most useful thing.
• Images: embedded rasters. Images that cannot be decoded, or that are linked from elsewhere, are counted rather than hidden, and a linked image's pixels are never fetched.
• Fonts: each with its embedding caveats stated: a subset is called a subset, a names-only face carries no bytes and "no embedding restriction" is not called a licence.
• Attachments.

Hidden words are reported first. Text painted underneath an opaque shape appears at the top of the report, with the hidden words shown. A black bar that does not remove the words under it is worth seeing rather than trusting.

Three hand-offs go straight into your design system: a font row installs its face, a mark goes to the Logos room of Design System Setup and the bar sends the whole scan there. None of them re-reads the document; each is built from what the passes already extracted.

Unpack reads up to 400 pages and 120 MB. Beyond that it reads what it can and says the rest is too long. A single mark over 4 MB is not sent to Design System Setup: download the SVG instead.

Read and extract

Read and extract (#/extract) gets the words, tables and pictures out of a PowerPoint deck (.pptx), a PDF or a Photoshop file (.psd), page by page and in reading order. Choose the Pages and set a Character budget; Continue reads on from where the budget stopped. Recognised text leaves out words read with low confidence.

A page that is a picture of its text offers Read the text on this device, which uses the on-device text reader. Each block read that way says which model read the words and how sure the model was. Take the result away with Copy text, or download it as Text, Markdown or JSON. Nothing is uploaded, and macros or scripts in the file never run. lolly read --extract gives the same result in the CLI.

Text

Text brings character exploration and text work into one tool. Its permanent tool ID is still text-helper, so existing links keep working.

• Characters & emoji browses the loaded brand font or searches emoji. Picking a character copies its exact Unicode sequence.
• Write & edit opens a persistent editor with syntax highlighting, selection actions and Undo. Highlighting is a display aid, not a code validator, and unknown languages stay readable as plain text. Snippet, Sandbox and code assets opened in Assets share the same highlighting. The editor and character grid use the active brand's font and interface colours.
• Find an action is a searchable chooser for editing, inspection, conversion and generation: the writing signals also shown in Verify and Assets, plain-language suggestions, de-identification with reviewable aliases, line and word comparison, regular-expression matching and replacement, JSON, YAML and TOML conversion, formatting and validation, encodings, checksums and ASCII lettering. Markdown can be previewed or downloaded as a document. QR codes stay in the QR tool.

Open a file or pick a text asset. Open in Text in Assets, Projects or a Verify text extract opens one too. Save adds the text to Assets, and to its source project when there is one. For text opened from one of your own assets the button reads Save changes, which updates that asset under the same ID so project references keep working; Save a copy in the same dialog keeps both versions. Save changes refuses to overwrite a source that changed after you opened the text. Text reads the complete file, up to 4 MiB, even where Assets shows a shortened preview. Drafts use normal saved-session storage and recovery, and edits do not put the document's contents into the address bar.

Check where a file came from

Verify & Inspect (#/verify) has a card in the Utilities tab and pages of its own: Verify It Yourself and Security & Verification.

PDF pages

Pages is the PDF page manager. Give it a 1-based expression such as 1-3,7,10- to reorder, rotate, extract or delete those pages. It can also append a second PDF or split comma-separated ranges into separate files. The page strip shows the selection before download. Pages keeps the main document's own metadata and adds none of Lolly's. It refuses encrypted PDFs and XFA forms rather than risk a partial document.

Pages also turns images into a PDF. Add photos, scans or screenshots, alone or with PDFs, and each image becomes one page in the same draggable sequence. Image page size sizes each page from the image's own resolution (Match each image, at 96 pixels per inch when the image gives none) or fits the image inside an A4 or US Letter page, with Image margin (mm) setting the margin.

Sign places a transparent signature on a chosen PDF page, in PDF-point coordinates: drag it on the page or enter the numbers. Initials and today's date are optional. A Content Credential is on by default and records the profile signer only when Use my details is on. A password lock is available at download time, but not together with that credential, because encrypting afterwards would break the credential. Sign makes a visible signed document with optional provenance, not a regulated e-signature or identity-verification service.

Audio and video

Trim cuts audio and video at exact in and out points on your device. Under Make it also takes a Poster frame (the picture at the start point) or a Contact sheet of frames across the range. For a clip it can:

• keep or change the container, or make a GIF of up to 15 seconds;
• Rotate the picture, Remove audio or Extract audio as M4A, Opus or WAV;
• re-encode to a Sharing size of 1080p, 720p or 480p, or aim for a Size limit (MB).

The result reports the real container and any codec the device could not handle.

Clean reduces voice noise with the on-device speech model and trims silent edges to the exact sample. It normalises to podcast, streaming or broadcast loudness and keeps the output under a -1 dBTP ceiling. Audio can leave as WAV, MP3, M4A or Opus. For a video, the picture is carried into the new container and the cleaned audio replaces its original track. The terminal path accepts WAV; other containers move to the browser tier or are refused by name.

Voice Recorder records a voice note with a live level meter and gentle coaching, then saves it as MP3.

Script audio (#/script) is a writing surface over on-device speech. Write or paste a script; Markdown is fine, because only the words are read, so code blocks and images drop out and links keep their text. Pick a voice, audition it, choose a speed and press Generate speech (or Ctrl or Cmd and Enter). Under the script sit the word count and an estimate of how long it takes to listen to.

• Shape the reading. Sound marks type marks into the script for you, such as more energy, a question, a breath, a pause, a slower sentence or a word spelled out in phonemes. Tips explains each one. Blend with mixes a second voice into the first, with one slider for the weight.
• Downloads and limits. The first run downloads the voice model once, with its size shown before you start. After that it runs on the device, and the script is never uploaded. A script past about five thousand characters gets a warning, not a refusal.
• Save the clip. Save to your uploads writes a WAV to your asset library, with the voice, speed and per-word timings kept on the record for captioning. It is marked as AI-generated and signed with a Content Credential in the WAV bytes (why).
• Fix a saved clip. Opening a saved clip in Script audio (#/script?asset=<id>) fills in its recipe. Save changes to this clip rewrites the same asset, so every document that uses the clip hears the fix; Save as new clip keeps both.

Where a device or browser has no speech support, the view says so in one sentence, and the Utilities tab shows no Script audio card.

Colour

Colour Lab (#/lab) reports on one colour in full. #/lab?c=<css colour> opens the report on any CSS colour, and the address follows what you pick, so a link reproduces the page you are looking at.

The report runs in five steps:

1. Set a colour with the app's tabbed colour picker, or from your brand's own swatches.
2. Plotted on a colour space: four charts, measured against a comparison target you choose.
3. Every notation: the same colour written for each space, ready to copy, with a clamped mark where a space is too narrow to hold the colour.
4. Tones and blends: a perceptually even ramp through the colour, and a blend to a second colour at a step count you set.
5. Displayable range and readability: the gamut verdict, then contrast scores. APCA comes first because it models polarity, and WCAG second because people still have to report WCAG scores. A foldable grid scores every brand colour against every other, with a colour-vision simulation on top.

signed by Lolly2026-10-23vector SVG337 paths~15k nodes343 groups4 images456 KBTự kiểm chứngTải tệp đã ký signed by Lolly2026-10-23vector SVG330 paths~15k nodes343 groups4 images456 KBTự kiểm chứngTải tệp đã ký

The colour is never collapsed to sRGB on the way in. color(display-p3 1 0 0) is described at its real chroma and gamut rather than flattened to #ff0000 and declared safe, and every swatch is painted from the value you wrote, so a wide-gamut display shows the real colour. A This screen panel at the foot reports the display the charts are judged against, read live and stored nowhere.

There is no export, canvas or CLI equivalent: copy the notations you need, or share the link.

Contrast checks a text and background pair, or every pairing in the design system palette, against WCAG 2.1 and APCA, and shows how it reads to colour-blind viewers. Palette Lab grows a palette from one seed colour, with harmony accents, perceptual OKLab ramps and readability badges. It exports DTCG tokens and CSV.

Spreadsheets

Spreadsheet (#/data) opens, reads and lightly edits a spreadsheet with no Excel, no LibreOffice and no internet. Drop or choose an .xlsx, .csv, .tsv or .json. A workbook with several sheets gets a tab bar, so you can switch sheets without picking the file again. Cells are editable in place, and Download as writes the grid as it stands, as CSV, Excel, JSON or TSV.

The grid draws only the rows in view and recycles them as you scroll, so a very large file stays responsive. The viewer reads up to 200,000 rows, and the reader stops at two million cells. Where either limit applies, the banner above the grid says how much you are looking at.

Spreadsheet shows values. A formula arrives as its last computed result, and styles, merged cells, charts and every sheet but the one you download are lost in the download. The banner says this before you edit anything. Use Spreadsheet to read a file, fix some numbers and pass the data on, not to round-trip a formatted workbook. Nothing is uploaded or kept: the file stays in the page until you leave.

Decks

Rebrand (#/rebrand) moves a PowerPoint or PDF deck onto your design system. It proposes each slide on the design system's own layouts, colours and typeface, and waits for you to review each suggestion before it opens the result in Design. Keep the design swaps only the theme, colours and fonts of a PowerPoint deck instead. Unlike the other views, Rebrand keeps each deck as a project on this device, so a review can wait until you come back. Rebrand a deck walks through it with a sample deck.

Theme swap is the tool for the quick case: it swaps the theme, colours and fonts in a PowerPoint deck on this device.

More utilities

• Screen Capture takes a screenshot or recording of your whole screen, a window or a browser tab. Drag on the canvas to crop; the export keeps the exact pixels you captured.
• URL Screenshot captures any web page at any scroll depth, with custom CSS. In a browser it needs the browser extension, which runs in Chromium-based browsers such as Chrome and Edge. The desktop app does the same job without the extension.
• Annotate marks up your own screenshot with arrows, boxes, numbered steps, callouts, a highlighter and a spotlight dim.
• Scan reads QR codes and barcodes on your device. It explains what a code carries (a link, Wi-Fi details, a contact or a product code) and copies first; opening, joining or adding is a separate, deliberate tap.
• Claim stamps your author, copyright and licence onto media you already made (an image, PDF, video or audio file) as Content Credentials, with a durable Imprint on images, and keeps any credential already in the file.
• Countdown is a focused countdown with a live progress ring. Click to pause.
• Jump makes a one-link landing page with your links, heading, portrait and colours. The whole page lives inside the link you share.
• Sandbox runs pasted HTML, CSS and JavaScript, or a JSX or TypeScript component, in a private offline sandbox, which suits previewing code from an AI assistant.
• Prompt Card typesets a long prompt into one compact image for a multimodal model, since image input often costs less than the same words as text.
• Rondocode lets you write music as code and hear it as you type, then export the song as audio, stems or MIDI with Content Credentials, or keep its source.

Related: Using Lolly for the tools and the gallery these sit beside. Search for reaching any utility by name. Export a file for what the tools themselves can write.