What is the LuvVoice skill?
The LuvVoice skill gives an AI agent a lightweight, file-based path to text-to-speech. Instead of connecting to a server first, the agent reads one markdown file, follows the instructions inside it, and calls the API directly.
Install one file
Drop the skill into the right client folder and the agent can discover the workflow immediately.
Teach the agent
The markdown file explains endpoints, parameters, defaults, and how to recover from common issues.
Return audio fast
Once loaded, the agent can generate speech, pick voices, and hand back a usable audio result.
How the skill workflow fits together
A skill is still a real integration surface, just a lighter one. The agent gets explicit instructions, then uses them to execute the task without you needing to run extra infrastructure.
Use the skill when simplicity matters most
curl or equivalent shell calls.The request can be natural language or an explicit command, depending on the client.
The file defines prerequisites, endpoint usage, default voice behavior, and response handling.
The agent makes the API call, receives the payload, and returns a playable link or file.
Installation stays client-specific, not workflow-specific
The workflow does not change across clients. What changes is only the file path and how that client chooses to invoke the skill.
Download the ready-made package
If you prefer not to curl the raw file directly, download the ZIP, extract it, and place the skill in the directory shown below.
~/.factory/skills/luvvoice-tts/SKILL.md/luvvoice-tts or auto-detected by Droidexport LUVVOICE_API_TOKEN=your_token_heremkdir -p ~/.factory/skills/luvvoice-tts
curl -o ~/.factory/skills/luvvoice-tts/SKILL.md \
https://raw.githubusercontent.com/luvvoice/luvvoice-mcp/main/.factory/skills/luvvoice-tts/SKILL.mdWhat the skill handles once installed
The point of the skill is not just setup convenience. It also gives the agent a reusable operating model for synthesis, discovery, and sensible defaults.
Text-to-Speech
Convert arbitrary text to MP3 output with voice, speed, pitch, and volume controls.
Voice Discovery
Filter the catalog by language or gender, then keep a shortlist of reliable starting voices.
Smart Defaults
When the user is vague, the agent can infer language, pick a natural default, and split long text safely.
Popular starting voices
The full voice catalog is still available at runtime. These are simply the voices people reach for first when testing the skill.
Keep the skill when you want lighter setup
The decision is less about features and more about integration posture. Both paths reach the same LuvVoice API, but they do it in different ways.
Choose the skill
Choose MCP
Decision rule
Ready to install it?
The practical sequence is still short: create a token, place the file, and let the agent follow the instructions from there.
Give your coding agent a voice in under a minute
Start with the skill if you want the lightest install path, and keep the MCP route nearby if your workflow later grows into a tool-based integration.