deAPI Skills

Agent skills for deAPI — ready-to-use bash scripts that let AI agents generate speech, clone voices, transcribe audio, and more.

Compatible with Claude Code, Cursor, Windsurf, and any agent platform that supports the community skills format. Up to 20x cheaper inference via decentralized GPU network.

Skills

Audio

Skill	Description	Scripts
deapi-audio	Text-to-speech, voice cloning, voice design, and audio transcription	text-to-speech.sh, voice-clone.sh, voice-design.sh, speech-to-text.sh

Setup

1. Get your API key

Sign up at deapi.ai (free $5 credit, no card required).

2. Set the key

export DEAPI_API_KEY="your_key_here"

Or let the agent store it in a config file on first use — it will ask for your key and write it to config.json inside the skill directory.

3. Install

git clone https://github.com/deapi-ai/skills.git
cp -r skills/deapi-audio ~/.claude/skills/

4. Use with any agent

Every skill is a standalone bash script. Call them directly:

# Text-to-speech
bash scripts/text-to-speech.sh --text "Hello world" --voice bf_emma

# Voice clone
bash scripts/voice-clone.sh --text "Hello in your voice" --ref-audio sample.mp3

# Voice design
bash scripts/voice-design.sh --text "Good morning" \
  --instruct "A warm male voice with a British accent"

# Transcribe audio
bash scripts/speech-to-text.sh --audio recording.mp3

Skill Format

Each skill follows the community skills standard:

deapi-audio/
├── SKILL.md              # Metadata + documentation (YAML frontmatter)
├── scripts/              # Executable bash scripts
│   ├── common.sh
│   ├── text-to-speech.sh
│   ├── voice-clone.sh
│   ├── voice-design.sh
│   └── speech-to-text.sh
└── references/           # Model docs, voice lists, API details
    ├── api.md
    └── voices.md

The SKILL.md contains a YAML frontmatter with name, description, and metadata, followed by usage documentation that the agent reads to understand how to use the skill.

License

MIT