Connect Kitta Audio TTS…dy with a Custom Skill

Install a beta WorkBuddy Skill that generates MP3 speech files through the Kitta Audio API.

Connect Kitta Audio TTS to WorkBuddy

The download retains the package name fish-audio-tts, the Kitta Audio display label, and the FISH_AUDIO_* compatibility variables. It does not change its name for this site. The explicit FISH_AUDIO_BASE_URL below selects the service; pair it with the matching Kitta Audio account key.

Tencent WorkBuddy can call external APIs through custom Skills. Use the Kitta Audio Skill to generate MP3 files from a WorkBuddy task.

This integration is in Beta because WorkBuddy's Skill import and permission UI can vary by desktop version. Do not configure Kitta Audio under Settings → Model: that screen expects a chat-model endpoint such as /chat/completions, not the Kitta Audio speech endpoint.

Before you start

You need:

A Kitta Audio API key with API credits
Python 3.9 or later available to WorkBuddy
Permission for the Skill to run a local script and access the network

Create a dedicated API key for WorkBuddy so it can be monitored and revoked independently.

Download the Skill

Download the Kitta Audio TTS Skill package and extract it. The package contains SKILL.md, skill.yml, a Python implementation, and a README.

WorkBuddy versions currently use one of these installation flows:

Open Skills, choose the import or local Skill action, and select the extracted folder.
If your version does not show an import action, create a task asking WorkBuddy to install the extracted folder as a custom Skill.

Review the files and the permissions shown by WorkBuddy before enabling the Skill.

Configure credentials

Set the key in the environment that starts WorkBuddy:

export FISH_AUDIO_BASE_URL="https://fishaudio.org/v1"
export FISH_AUDIO_API_KEY="YOUR_FISH_AUDIO_API_KEY"
export FISH_AUDIO_VOICE_ID="00a1b221-6137-4b73-ad62-b0cbce134167"
export FISH_AUDIO_MODEL="fishaudio-s21pro-flash"

On macOS, environment variables set in a terminal are not automatically available to an app launched from Finder. Follow your WorkBuddy version's local-secret or environment configuration flow, or start the app from an environment that contains these variables.

Never paste the API key into a WorkBuddy conversation or save it in the Skill folder.

Test the Skill

Start a new WorkBuddy task and ask:

Use the fish-audio-tts Skill to create an MP3 file that says:
"Hello, this is a Kitta Audio test."
Save it as fish-audio-test.mp3.

Confirm that WorkBuddy reports the output path and that the MP3 plays. Then replace the test Voice ID with your preferred Kitta Audio Voice ID.

Troubleshooting

Skill is not discovered: confirm that the extracted folder contains SKILL.md at its root, then re-import or restart WorkBuddy.
Python or terminal permission is denied: enable only the local execution permission required by this Skill.
API key is missing: expose FISH_AUDIO_API_KEY to the WorkBuddy process; do not add it to the Skill files.
401/402: replace the API key or add API credits.
No MP3 is created: inspect the task output for the HTTP status and response body, and confirm that the output directory is writable.

Beta support boundary

The included script and Kitta Audio API request are testable independently. The exact Skill import screen and secret-loading behavior must still be verified against the WorkBuddy desktop version you use. If installation fails, include the WorkBuddy version, operating system, and the visible error message when contacting support.

WorkBuddy Setup (Beta)