Guide · Agent skill
How to use the WaveTone Audio Workflow Control agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
This skill teaches an AI agent to prepare and launch WaveTone 2.61, a Windows audio transcription program, using a small JSON project file. It is aimed at musicians and transcribers who already use WaveTone on Windows, and at developers who want a WaveTone step in an automated pipeline. After this guide you will know how to install the skill, ask an agent for useful requests, and check the requirements and safety notes before running it.
WaveTone Audio Workflow ControlHKUDS/CLI-Anything · Apache-2.0 Download the skill Quick start
- 1Check requirements. Confirm you have a Windows host with WaveTone 2.61 extracted locally, and that you have a legitimately obtained copy.
- 2Install the cli-anything-wavetone tool. Install the command-line tool separately, as it is not included in the skill files.
- 3Set the environment variable. Set WAVETONE_EXE to wavetone.exe or WAVETONE_HOME to the extracted WaveTone directory if the default portable path is not used.
- 4Install the skill. Unzip the downloaded ZIP into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 5Ask the agent for a task. Describe the audio file, tempo, meter and labels in plain language and mention WaveTone.
- 6Run the doctor check. Let the agent run wavetone doctor to confirm WaveTone is found.
- 7Launch and analyse in WaveTone. Let the agent launch WaveTone with your project, then run the analysis and save WFD, MIDI, text or WAVE outputs inside WaveTone.
What the skill makes the agent do
Agent skills are instruction packages that an AI agent loads when a task matches them. This skill, named cli-anything-wavetone, tells the agent how to work with WaveTone 2.61 workflows. The agent can create a JSON project file that records a source audio file, tempo, meter, labels and intended analysis settings. It can also probe audio metadata, check that WaveTone is installed, and launch the real WaveTone application.
The JSON manifest is only a plan. It is not a WaveTone WFD file. The analysis and saving of WFD, MIDI, text and WAVE outputs is still done by WaveTone itself, so the agent does not transcribe audio on its own.
What you need before installing
The skill requires a Windows host and WaveTone 2.61 extracted locally. It does not work on macOS or Linux. You also need the cli-anything-wavetone command-line tool, which is installed separately and is not included in the files reviewed for this skill.
Set the environment variable WAVETONE_EXE to the path of wavetone.exe, or WAVETONE_HOME to the extracted WaveTone directory, when the default portable path is not used. The skill does not provide a WaveTone copy, so you must already have a legitimately obtained one.
- Windows host
- WaveTone 2.61 extracted locally
- WAVETONE_EXE or WAVETONE_HOME environment variable set when needed
- cli-anything-wavetone command-line tool installed separately
How to install the skill
Download the ZIP from the skill page on this site and unzip it. For a personal install, place the skill folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.
In the Claude apps you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/
- Claude apps: Settings → Capabilities → Skills, then upload
How to trigger the skill with natural requests
You do not need to type a command to use the skill. Describe the task in plain language and mention WaveTone, so the agent recognises a match. The skill's own description covers preparing and inspecting WaveTone 2.61 transcription workflows and launching the WaveTone executable.
Good requests name the audio file, the tempo or meter, and what you want saved. Say whether you want the agent to only prepare the project or also open WaveTone.
- Create a WaveTone project for input.wav at 120 BPM in 4/4 and add a label called chorus at 64 seconds.
- Check whether WaveTone is installed correctly on this Windows machine.
- Probe the metadata of song.wav before I open it in WaveTone.
- Launch WaveTone with my project file so I can run the analysis.
Core commands the agent uses
The agent runs commands from the cli-anything-wavetone tool. All commands support the top-level --json flag. The main groups are project (create manifests, set tempo, add labels, update intended analysis settings and attach a WFD saved from WaveTone), audio (probe metadata), wavetone (run doctor, list supported formats or launch the GUI), session (record lightweight event logs for multi-step workflows) and defaults (show default analysis settings).
cli-anything-wavetone --json wavetone doctor
cli-anything-wavetone --json project new input.wav -o project.wt.json
cli-anything-wavetone --project project.wt.json --json audio probe
cli-anything-wavetone --project project.wt.json --json project set-tempo --bpm 120 --meter 4/4
cli-anything-wavetone --project project.wt.json --json project add-label chorus --time 64.0
cli-anything-wavetone --project project.wt.json --json wavetone launchAutomated checks and pipelines
For automated checks, the source shows a launch command that starts the real backend without leaving a persistent GUI process behind. This confirms that WaveTone can be started in a pipeline step.
Keep in mind that the agent prepares the project and launches WaveTone, but a person or your pipeline still needs to run the analysis inside WaveTone and save the outputs.
cli-anything-wavetone --json wavetone launch input.wav --wait 1 --terminateSafety notes: scripts and network
This skill runs scripts. It launches the real WaveTone executable, so review the commands the agent proposes before it runs them, especially launch commands. The skill does not use the network, according to its listing.
Only point WAVETONE_EXE or WAVETONE_HOME at a WaveTone 2.61 install you trust and that you have legitimately obtained.
- Runs scripts: yes, including launching wavetone.exe
- Uses network: no
- Check commands before the agent runs them on a machine with important files
Tips for getting good results
Run the doctor command first to confirm the environment is set up. Keep the project file next to the audio it refers to so paths stay simple. Remember that the intended analysis settings in the manifest are a plan, so check them inside WaveTone before you save final outputs. Use the defaults command to see the default analysis settings before you change them.
- Run wavetone doctor before any other step
- Keep project.wt.json and the audio file in the same folder
- Verify analysis settings inside WaveTone
- Use --json for output the agent can parse
What people use it for
- Preparing a project file with tempo, meter and labels for a transcription session in WaveTone.
- Probing audio metadata on a set of recordings before opening them in WaveTone.
- Checking that a Windows machine has WaveTone installed correctly before a batch of work.
- Adding a Windows WaveTone launch step to an automated audio-transcription pipeline.
Frequently asked questions
How do I install the cli-anything-wavetone skill?
Download the ZIP from the skill page and unzip it into ~/.claude/skills/ for a personal install, or into a project's .claude/skills/ folder. In the Claude apps, you can upload it under Settings → Capabilities → Skills.
What do I need to use the WaveTone skill?
You need a Windows host, WaveTone 2.61 extracted locally, the WAVETONE_EXE or WAVETONE_HOME environment variable set when the default path is not used, and the cli-anything-wavetone command-line tool installed separately.
Is the WaveTone agent skill free?
The skill is licensed under Apache-2.0. WaveTone 2.61 is a separate program, and you need your own legitimately obtained copy, which the skill does not provide.
Does this skill work on Mac or Linux?
No. The skill requires a Windows host and the WaveTone executable, so it is not for macOS or Linux users.
Does the agent do the transcription itself?
No. The JSON manifest is only a plan. WaveTone performs the analysis and saves WFD, MIDI, text and WAVE outputs.
Why is wavetone.exe not found when I launch?
The skill expects WAVETONE_EXE set to wavetone.exe or WAVETONE_HOME set to the extracted WaveTone directory when the default portable path is not used. Run the doctor command to check the setup.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you already run WaveTone 2.61 on Windows and want an agent to prepare project files and launch the program, or if you are building a Windows transcription pipeline. Pick something else if you use macOS or Linux, do not have WaveTone, or expect the agent to perform the transcription itself.
Download the skill