Agent Skills

liveavatar-debug

A collection of skills to help coding agents onboard to LiveAvatar.

Installation
CLI
npx skills add https://github.com/heygen-com/liveavatar-agent-skills --skill liveavatar-debug

Install this skill with the CLI and start using the SKILL.md workflow in your workspace.

Last updated 4/22/2026

LiveAvatar Agent Skills

Reusable skills for AI coding agents integrating with LiveAvatar. These skills provide procedural knowledge that helps agents build LiveAvatar integrations correctly on the first attempt.

Available Skills

Skill Description
liveavatar-integrate End-to-end integration builder — assesses your existing stack, recommends the optimal path (Embed / FULL / LITE), and guides implementation step by step.
liveavatar-debug Symptom-based troubleshooting for silent avatars, garbled audio, auth errors, and more.
liveavatar-feedback Collects user feedback on their LiveAvatar integration experience and sends it to the LiveAvatar team. Triggers after implementation, on frustration, or on explicit request.

Installation

# Install all skills globally
npx skills add heygen-com/liveavatar-agent-skills -a claude-code -g

# Or install to current project only
npx skills add heygen-com/liveavatar-agent-skills -a claude-code

# Install a specific skill only
npx skills add heygen-com/liveavatar-agent-skills --skill liveavatar-integrate

This works with Claude Code, Cursor, Codex, and other agents.

Option 2: Manual Installation

git clone https://github.com/heygen-com/liveavatar-agent-skills.git

# Symlink all skills to personal skills directory (available in all projects)
for skill in liveavatar-agent-skills/skills/*/; do
  ln -s "$(pwd)/$skill" ~/.claude/skills/$(basename "$skill")
done

Skills activate automatically when agents detect relevant tasks (e.g., "add a LiveAvatar avatar", "build a conversational avatar", "integrate LiveAvatar").

Skill Structure

Each skill follows the Agent Skills format:

skill-name/
├── SKILL.md          # Agent instructions with behavioral guidance
└── references/       # Supporting documentation (API details, code examples)

Quick Reference

Task Skill
Build a new LiveAvatar integration liveavatar-integrate
Put an avatar on a page (no code) liveavatar-integrate → Embed pathway
Build a conversational avatar liveavatar-integrate → FULL Mode pathway
Use your own LLM / TTS / full pipeline liveavatar-integrate → FULL or LITE pathway
Connect ElevenLabs Agents to an avatar liveavatar-integrate → LITE + ElevenLabs pathway
Debug a silent or broken avatar liveavatar-debug
Share feedback about the integration experience liveavatar-feedback

Example Prompts

"Add a LiveAvatar avatar to my landing page"

"Build a conversational AI avatar for customer support"

"I have my own LLM — integrate it with LiveAvatar"

"Set up LiveAvatar LITE Mode with my Pipecat pipeline"

"My LiveAvatar avatar is silent, help me debug"

"Create a sandbox session to test LiveAvatar"

"I want to give feedback about the LiveAvatar integration"

"This integration is frustrating, let me report this"

Requirements

  • LiveAvatar API key (get one at LiveAvatar Dashboard)
  • Claude Code CLI (or any skills-compatible agent)

API Reference

All skills use the LiveAvatar API:

  • Base URL: https://api.liveavatar.com
  • Authentication: X-API-KEY header (backend), Bearer <session_token> (session ops)
  • API Versions: v1 (sessions, avatars, voices, contexts), v2 (embeds)

Design Philosophy

  • Pit of success: The correct path is the easiest path. Always default to the simplest integration.
  • Gotcha-driven: Lead with what breaks. Silent failures (missing context, wrong audio format) are called out before agents hit them.
  • Backend/frontend split: Every code block is labeled with where it runs and which auth to use.
  • Progressive disclosure: SKILL.md has guidance and structure. References have API specifics and code examples.

Contributing

Contributions welcome! See CONTRIBUTING.md for guidelines. When adding or modifying skills:

  1. Lead with gotchas — what breaks if you do it wrong?
  2. Be explicit about backend vs frontend and which auth header to use
  3. Include trigger phrases in the skill description
  4. Keep SKILL.md under 500 lines — move details to references/

License

MIT