About this toolkit
An independent set of guides and browser tools for people writing prompts and managing files around AI music generation.
Who this is for
People who have generated a few songs, noticed that results vary enormously with wording, and want to get deliberate about it. Hobbyists shipping to friends, creators scoring their own videos, and writers who treat the prompt as a craft object rather than a magic spell.
It is also for anyone who has ended up with sixty untitled files in a Downloads folder and wants a tidier way to pull a track, its artwork and its lyrics out in one pass.
- You want vocabulary — the actual words for instruments, vocal delivery, production character and tempo feel.
- You want a method for testing changes rather than a list of magic phrases.
- You want file tools that run in your browser and do not ask you to sign up for anything.
What is here
| Area | What you get |
|---|---|
| Guide | Sixteen topic hubs and a forty-lesson course covering prompt anatomy, genre, vocals, instruments, mood, structure and production |
| Styles | A searchable library of prompt-ready style lines by category, mood and tag |
| Tools | Sixteen browser tools for building, cleaning, comparing and formatting prompts, plus downloading and analysing finished tracks |
| Articles | Longer field notes on workflow, formats and troubleshooting |
| Lab | Structured prompt experiments with published designs you can rerun |
How it is built
Plain HTML, CSS and JavaScript. No framework, no build step beyond a small static assembler, no client-side router. Pages are files; the site works with JavaScript disabled for everything except the tools themselves.
The tools are local-first by design. Text processing — building, cleaning, diffing, formatting, normalising, timing — happens entirely in your browser and never leaves the machine. Audio decoding, transcoding and ZIP packaging also run client-side, using your CPU rather than a server.
- No accounts. Nothing to register, nothing to log into.
- No analytics or trackers. No third-party scripts of any kind.
- No CDNs. Every asset is served from this domain, so your browsing is not announced to third parties.
- One network exception: when you paste a public song, playlist or profile URL, the site's own proxy fetches that public page so your browser can read it. See the privacy page for the detail.
Editorial standards
Every page here is written from hands-on use. Where we are describing something we tested, we say so and publish the design. Where we are describing a convention or a habit, we say that too.
We do not publish invented measurements. Lab studies ship as reusable frameworks with template tables, not as charts of numbers you cannot verify. We do not document plan limits, pricing or interface details that change without notice — the platform's own documentation is authoritative for those, and a stale page is worse than no page.
Open-source credits
Two third-party libraries are bundled with the tools, both served from this domain rather than a CDN and both used as distributed, without modification. Licence texts ship alongside the library files. Everything else on the site — markup, styles, scripts, datasets and copy — is original work by this project.
lamejs is LGPL-licensed: it is included as a separate, unmodified library file rather than compiled into our code, and can be replaced by a user with a compatible build.
| Library | Licence | Used for |
|---|---|---|
| JSZip | MIT | Building ZIP archives in the browser for multi-file and playlist downloads |
| lamejs | LGPL | Encoding MP3 audio client-side from decoded audio data |
Independence and disclaimer
Nothing here is official documentation, and nothing here is legal advice. Model behaviour changes, interfaces change, and terms change; verify anything that matters against the source. The tools read public pages only, and you remain responsible for having the rights to whatever you download or publish.
See the terms for the full position, and the privacy page for what does and does not leave your browser.