agentleFS
heyderekjheyderekj/dinky/site/llms.txt

dinky / site

llms.txt519 starsChanged 7 months agoMITView on GitHub

A tiny macOS compression app for images, videos, audio, and PDFs. Converts stills to WebP, AVIF, lossless PNG, or HEIC; exports video to smaller MP4 (H.264 or HEVC) with optional FPS cap; cross-converts AAC, ALAC, WAV, AIFF, FLAC, and MP3; shrinks PDFs. Drag and drop, get smaller files back. Free, open source, and entirely local — files never leave your machine.

  • Installs packages

We split this llms.txt into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Update checker: polls GitHub Releases on launch, shows banner if newer version exists

    line 26
  2. Distinctive:

    No cloud, no accounts, no in-app telemetry SDKs

    line 41

≈ “Common” is an estimate: at least 1 in 5 of the 3,225 llms.txt files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
llms.txt
# Dinky

> A tiny macOS compression app for images, videos, audio, and PDFs. Converts stills to WebP, AVIF, lossless PNG, or HEIC; exports video to smaller MP4 (H.264 or HEVC) with optional FPS cap; cross-converts AAC, ALAC, WAV, AIFF, FLAC, and MP3; shrinks PDFs. Drag and drop, get smaller files back. Free, open source, and entirely local — files never leave your machine.

## What it does

- **Images** — JPG, PNG, WebP, AVIF, HEIC/HEIF, TIFF, BMP → WebP, AVIF, lossless PNG, or HEIC
- **Videos** — common formats (e.g. MP4, MOV) → MP4 with H.264 or HEVC and quality presets; optional FPS cap (lower output frame rate when the source is higher)
- **Audio** — AAC (M4A), ALAC, WAV, AIFF, FLAC, MP3 cross-convert via macOS `afconvert`; MP3 encode via bundled **lame**
- **PDFs** — **Flatten (default)** rasterizes pages (PDFKit + JPEG) for reliable size wins; **Preserve (best-effort)** uses bundled **qpdf** (with lossless stream coalescing + resource cleanup) then PDFKit — only keeps output when smaller. **Scan OCR (optional)**: Vision text layer on scan-like PDFs before flatten/preserve. **Target file size**: set a ceiling (e.g. 10 MB) and Dinky steps down quality tiers until the output fits. **Image downsampling**: opt-in mixed-mode rasterizes image-heavy pages at 144 DPI while keeping text/link pages selectable — high-impact on 300 DPI scan PDFs
- Drag and drop or file picker; batch compression with live results
- **URL drop** — drag or paste a direct `http(s)` link to an image, video, audio file, or PDF; Dinky downloads (max 500 MB), compresses, and cleans up the temp file
- **Clipboard Compress** — system-wide hotkey (default ⌘⇧V) works even when Dinky isn't focused; pastes file URLs, raw image bytes, or media URLs from the clipboard
- **Custom keyboard shortcuts** — rebind Open Files, Clipboard Compress, Compress Now, Clear, and Delete in Settings → Shortcuts
- **Launch at login** — opt-in via SMAppService so Dinky and any watch folders are ready at login
- **Localized** — functional UI follows the system language for 12 Tier-1 macOS locales (de, es, fr, it, ja, ko, nl, pt-BR, ru, tr, zh-Hans, zh-Hant); brand voice and playful copy stay in English
- **Originals** — keep, move to Trash, or move to a Backup folder (global setting; watch folders can override it)
- **Batch speed** — Fast (one at a time), Faster (up to three parallel), or Fastest (up to eight)
- Max width resizing (common web presets or custom value)
- Max file size targeting via binary search on quality (images); step-down quality targeting for PDFs
- Smart quality: detects photos vs graphics (UI, illustrations, logos, screenshots) via Vision framework + EXIF, adjusts quality per content type
- Finder Quick Actions — compress without opening the app
- Manual mode: drop files, then right-click each one to pick format individually
- Skip threshold: skip files below a minimum savings target — Off, 2%, 5%, or 10%
- Completion sound scales with savings: Tink (< 100 KB), Pop (< 1 MB), Glass (< 5 MB), Hero (5 MB+)
- Update checker: polls GitHub Releases on launch, shows banner if newer version exists

## Version line

- **1.x** (from 1.0): images only. **2.0+**: images, videos, and PDFs. **2.10+**: adds audio conversion and video FPS cap. Older 1.x binaries remain on GitHub Releases for archival use.

## Technical details

- Apple Silicon (M-series Mac) only for distributed 2.x builds; Intel-based Macs are not supported
- macOS 15 Sequoia or later (liquid glass UI on macOS 26 Tahoe; frosted material fallback on Sequoia)
- SwiftUI + AppKit; the app uses an in-tree Swift package for shared image code + optional `dinky` CLI (not a remote SPM dependency)
- 35 MB installed
- Image engines bundled: cwebp (Google), avifenc (AOM), oxipng (Rust/MIT)
- Audio: cross-convert via macOS **`afconvert`**; MP3 encode via bundled **`lame`**
- PDF: flatten via PDFKit + ImageIO; preserve path via bundled **qpdf** + PDFKit; video via AVFoundation export
- No cloud, no accounts, no in-app telemetry SDKs
- Open source: MIT license

## Network behavior

- **Update check:** at most once per 24 hours after the first successful check, the app requests GitHub’s public API for the latest release (`Dinky/Services/UpdateChecker.swift`). No file contents are sent.
- **Crash diagnostics:** optional in-app toggle; Apple’s MetricKit delivers payloads on-device. Nothing is transmitted until the user chooses email or a prefilled GitHub issue (`Dinky/Services/DiagnosticsReporter.swift`).
- **URL Drop / Clipboard Compress:** HTTPS download only for a direct file URL you paste or drag; no background pre-fetch. Marketing site **dinkyfiles.com** uses Fathom for page analytics only; the app does not.

## Comparison

- vs **ImageOptim**: ImageOptim is same-format (lossless) — a JPEG stays a JPEG. Dinky **converts** stills to WebP/AVIF/HEIC/PNG for larger wins; use ImageOptim when you must keep the original extension.
- vs **TinyPNG**: TinyPNG uploads images to their servers. Dinky is 100% local.
- vs **Squoosh**: Browser-based, one image at a time. Dinky is native, batch, offline.
- vs **Optimage**: Similar feature set but Optimage costs $15 and weighs 62 MB. Dinky is free and 35 MB.
- vs **HandBrake**: HandBrake is built for deep video encoding (tracks, chapters, many settings). Dinky is drop-and-done for smaller MP4s and lives beside images, audio, and PDFs in one app.
- vs **QuickTime**: QuickTime ships with macOS for playback and simple one-file export. Dinky adds named H.264/HEVC presets, FPS cap, batch folders, watch folders, audio cross-converts, still-image plus PDF compression in the same app.
- vs **Apple Compressor**: Compressor is for Final Cut and pro encoding batches. Dinky is a lightweight drop zone when you only need a smaller MP4 plus images, audio, and PDFs in the same MIT app.
- vs **Permute**: Permute is a broad Mac media converter. Dinky is compression-first with open source PDF and still-image pipelines alongside video and audio.
- vs **FFmpeg**: FFmpeg is CLI-first with unlimited flexibility. Dinky wraps common shrink tasks in a native GUI and adds PDFs, audio, and stills—no shell flags.
- vs **Adobe Acrobat**: Acrobat is a full subscription PDF suite and a large install. Dinky targets straight compression—flatten, preserve, target size, optional scan OCR—in a ~35 MB free app.
- vs **Preview**: Preview handles still exports and smaller PDFs via Quartz filters for one-off saves. Dinky adds batch, watch folders, target size, WebP/AVIF/HEIC conversion for images, flatten/preserve PDF modes, and mixed media in one workflow.
- vs **PDF Squeezer**: Both are Mac-native PDF compressors; Dinky is MIT-licensed and bundles image + MP4 + audio compression in the same app.
- vs **Smallpdf**: Smallpdf is browser-first with uploads. Dinky keeps PDF, image, video, and audio compression on disk.
- vs **Clop**: Clop is menu-bar and clipboard–first with a Pro tier for some automation. Dinky is a free MIT window–first drop zone with watch folders included and deeper PDF tooling in a smaller install.
- vs **Picmal**: Picmal is a paid Mac app for images, video, PDFs, and audio with a large footprint. Dinky is free, open source, ~35 MB, and covers images, video, PDFs, and audio—with a vastly smaller bundle.
- vs **Smol**: Smol is a $29 menu-bar app (macOS 13+, Apple Silicon only) for images, PDFs, and video—Smart Clipboard, batch folder drops, no watch-folder automation, no audio. Dinky is free MIT, ~35 MB, adds audio and watch folders, and offers CLI automation—without RAW or AI upscale today.

## Links

- Website: https://dinkyfiles.com
- Comparisons index: https://dinkyfiles.com/compare/
- Comparisons (SEO): https://dinkyfiles.com/compare/imageoptim/, https://dinkyfiles.com/compare/optimage/, https://dinkyfiles.com/compare/tinypng/, https://dinkyfiles.com/compare/squoosh/, https://dinkyfiles.com/compare/clop/, https://dinkyfiles.com/compare/picmal/, https://dinkyfiles.com/compare/smol/, https://dinkyfiles.com/compare/handbrake/, https://dinkyfiles.com/compare/compressor/, https://dinkyfiles.com/compare/permute/, https://dinkyfiles.com/compare/ffmpeg/, https://dinkyfiles.com/compare/quicktime/, https://dinkyfiles.com/compare/adobe-acrobat/, https://dinkyfiles.com/compare/preview/, https://dinkyfiles.com/compare/pdf-expert/, https://dinkyfiles.com/compare/pdf-squeezer/, https://dinkyfiles.com/compare/smallpdf/, https://dinkyfiles.com/compare/xld/, https://dinkyfiles.com/compare/freac/
- Support: help@dinkyfiles.com
- GitHub: https://github.com/heyderekj/dinky
- Latest release: https://github.com/heyderekj/dinky/releases/latest
- Download v2.15.0: https://github.com/heyderekj/dinky/releases/download/v2.15.0/Dinky-2.15.0.dmg
- Homebrew (optional): `brew tap heyderekj/dinky https://github.com/heyderekj/dinky` then `brew install --cask dinky`

## Author

Derek Castelli — freelance web designer (Webflow, Figma, Cursor)
- Website: https://heyderekj.com
- GitHub: https://github.com/heyderekj

## Agent integration

This website is a static marketing page for the Dinky macOS app. There is **no public HTTP API**, **no OAuth/OIDC**, **no protected API resource**, and **no remote MCP server** on dinkyfiles.com.

- Machine-readable discovery: `https://dinkyfiles.com/openapi.yaml` (empty paths for the *site*; see description for the local product), `https://dinkyfiles.com/.well-known/api-catalog`, `https://dinkyfiles.com/.well-known/agent-skills/index.json`, and this file (`https://dinkyfiles.com/llms.txt`).
- **Local automation (images):** the open-source app repo ships a `dinky` **CLI** and optional **loopback-only**, token-protected `dinky serve` (build from `DinkyCoreImage/` on a Mac; listens on `127.0.0.1`/`[::1]` only, POSTs need `Authorization: Bearer <token>` and `Content-Type: application/json`, browser `Origin` requests are refused). Same JSON contract for `--json` and `POST /v1/compress`. Documented in-repo: `docs/local-cli.md` (schema `dinky.image.compress/1.0.0`). **No** upload to dinkyfiles.com; paths are explicit and on-device.
- **WebMCP** (Chromium): when supported, the homepage registers in-browser tools via `navigator.modelContext.registerTool()` — e.g. scroll to FAQ, open install help, copy the quarantine Terminal command, open GitHub Releases.
- Automation outside the browser: use **GitHub** (source, issues, releases) and direct **DMG download URLs** from the Releases page for the GUI app. For image compression from Terminal/agents, use the **local** `dinky` tool from the repository — not this website.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.