FAQ

Digr lets you search the sample library already on your computer by asking for what you want in plain words, in the same chat where you talk to Claude. Digr isn't a standalone app: it works with Claude Desktop. Ask for "808 kicks", "vocal loops" or "shakers between 170 and 178 BPM", and it looks through the names of every file and folder in your library and brings back the best matches. It can also work out a sample's tempo and key, read MIDI files, and copy, sort or rename files for you. Before it changes anything, it shows you exactly what it's about to do.

Yes, that's who it's built for. Most libraries grow faster than anyone can organise them. You download a pack for one vocal and two thousand samples come with it. Digr doesn't ask you to tidy anything first. There's nothing to import and no database to build, because it reads your folders as they are each time you search.

No commands and no extra app. You ask Claude in the chat you already use, the way you'd say it out loud: "any hi-hat loops at 140?", "find me a break beat between 125 and 135". Claude turns that into searches Digr runs on your library, then shows you what it found. If you can type a question, you can use Digr.

Only when those words are in the name. Digr searches the words in your file and folder names; it doesn't listen to the audio to decide whether something sounds dark or warm. Plenty of packs do put words like "dark", "warm" or "dirty" in their file names, and when they do, Digr finds them. When they don't, ask for the kind of sound (pads, drones, basses) and add the tempo or key if you know it. And if nothing matches every word you typed, Digr shows you the closest matches and tells you which word each one was missing, instead of just saying "no results".

Digr finds what your files are called, so it's only as good as the names — but it's forgiving about how they're written. "Break" finds "breaks", "hi-hat" finds "hihat" and "hat", "perc" finds "percussion", and drum-machine shorthand like BD, SD and HH finds kicks, snares and hi-hats. Search doesn't add tags of its own, so what you see is what your files are called. Results are ranked across your whole library: files that match every word come first, then the closest near-misses, each labelled with the word it missed. The best test is a corner of your library you know well. Search it and see whether Digr brings back what you'd expect.

No. Digr runs locally: your files aren't uploaded, and nothing is used to train AI. Digr reads and analyses your samples on your own drive, and the audio never leaves your computer, which matters if you work on client, label or unreleased material. Your library stays local and private. The words you type, and the file names Digr finds, pass through your Claude chat like any other message; the sounds themselves never do.

No. Nothing you do in Digr is used to train AI, and your audio is never uploaded for anyone to collect. You don't have to take our word for it: Digr's code is source-available on GitHub, which means anyone can read exactly what it does with your files.

Only when you ask, and never without showing you first. Searching and analysing are read-only. Your files stay where they are, named as they are. When you ask Digr to copy, move, sort or rename, it shows you a preview of what will happen and waits for your go-ahead. Copying leaves your originals untouched, and files are only moved if you ask for a move. There's no delete command at all.

Yes. Every rename is logged, and you can ask Digr to undo the most recent batch. It shows you what it will put back and asks you to confirm. It remembers your last 50 batches, so you can step back one at a time. If a file has been moved since, or its old name is now taken, Digr skips it and tells you rather than guessing. Two more safeguards: adding BPM or key to file names is off unless you ask for it, and if a file already has a tempo or key in its name, Digr keeps it and doesn't add a second one.

Yes. There's no code to write and no terminal to open. The simplest way is to download the Digr extension file from our GitHub releases page, then in Claude Desktop go to Settings, then Extensions, and install it. Claude does the rest. There are also one-click installers for Mac and Windows if you'd rather use those. On a Mac, the first time you open the installer you may need to right-click it and choose Open, which is macOS's normal check for anything downloaded.

Yes. Anthropic makes desktop extensions available to everyone who uses Claude Desktop, and Digr installs as one. Claude's usage limits still apply, and those are set by Anthropic, not by us. Digr works best with Claude's Sonnet and Opus models, which follow its multi-step jobs (search, then copy the results, say) most reliably.

Digr runs on Mac and Windows. It works with Claude Desktop, which sets it up automatically, and it's also been tested with Claude Code. Other MCP-compatible assistants (MCP is the standard Claude uses to connect tools like Digr) should work too, but we haven't tested them, so we won't promise it. As for DAWs, any one, or none: Digr searches your sample library, not your DAW, so it doesn't matter what you produce in.

When Digr starts, it looks where samples usually live: Ableton Live 12's libraries, your Splice downloads, and folders called "Samples", "Audio Samples" or "Sample Library" on any connected drive. If you use an older version of Live or another DAW, or keep samples somewhere else, just add the folder. Copy its location from Finder or File Explorer, paste it into the chat and ask Claude to add it. Digr remembers it for every session after that. You can remove a folder the same way, and it stays removed, even one Digr found by itself.

WAV, AIFF, MP3, FLAC and OGG audio, plus MIDI files, whether the file extension is written in capitals or not. It doesn't handle M4A or AAC files. One thing for Ableton users: Digr can find and copy the samples in Ableton's own Packs, but it can't work out the tempo or key of many of them. There's a separate question on that below.

It depends on whether the file name already tells Digr. If a sample's name includes its tempo or key (say "_174_" or "_Am"), Digr uses that, because the producer who made the sound knows best. When the name doesn't say, Digr works it out from the audio, and that's a hard thing to get right. Tempo readings can land on half or double the real speed. Key detection can't tell major from minor, and therefore sometimes picks a strong note that isn't the key. So anything Digr works out from the audio is marked as an estimate, and very short sounds like one-shots are reported as having no tempo rather than given a made-up number. Getting detection as accurate as possible is a difficult task and an ongoing process.

Many of the AIFF files in Ableton's official Packs are saved in Ableton's own compressed format, which only Ableton can play. Digr can find, list and copy them like any other sample, and they load in Live as normal. What it can't do is analyse their audio for tempo or key. If you ask, Digr tells you plainly that the file uses Ableton's format rather than throwing an error at you. In a BPM search, files that have the tempo in their name still show up.

Yes. You can search for MIDI files just like audio, and ask Claude to read one: Digr shows you its tempo, time signature and notes, laid out by bar and beat. If a file doesn't have a tempo or time signature saved in it, Digr says so and tells you the standard default it's assuming (120 BPM and 4/4), rather than presenting it as a real reading. It doesn't work out the key of a MIDI file.

Digr is made by Sentinel Protocol, a small UK company, and built by a producer who was fed up with searching through his own sample library. The code is public on GitHub, so you can see it's real.

Updates aren't automatic. When a new version comes out, download it from the Digr releases page on GitHub and install it the same way you did the first time. Your library folders and settings are kept, because they're stored separately from Digr itself.

Want first word on what we build next?

More tools are coming.

You can Subscribe on our webpage and you'll be first to hear about them.

https://sentinelprotocol.co.uk/digr/#subscribe

That's all we'll send; no noise.