avatar-content-engine

Portable AgentSkills skill for running an AI-avatar short-form content engine — script, image, and lip-synced video for TikTok, Reels & Shorts. Works in OpenClaw & Claude Code.

Avatar Content Engine — Agent Skill

A portable AgentSkills skill for running an AI-avatar short-form content engine — script, image, and video production for social-video avatars across TikTok, Reels, Shorts, and Facebook.

It encodes a full production methodology: an emotion-first content philosophy, a structured 16-second two-clip script contract, ten toggleable virality systems, a photographic-language approach to consistent avatar portraits, and a voice/video architecture for lip-synced multi-clip sequences.

The skill is a reusable template — the example avatars (Nova, Remy, Kai, Sable) and the [BRAND] slots are meant to be renamed and filled with your own product.

Install

OpenClaw

# From a Git URL
openclaw skills install https://github.com/<you>/avatar-content-engine

# Or drop it into your workspace manually
mkdir -p <workspace>/skills/avatar-content-engine
cp SKILL.md <workspace>/skills/avatar-content-engine/SKILL.md

Start a new session (or openclaw gateway restart) so the skill snapshot refreshes, then invoke it with /avatar-content-engine or let it auto-trigger.

Claude Code

mkdir -p .claude/skills/avatar-content-engine
cp SKILL.md .claude/skills/avatar-content-engine/SKILL.md

Compatibility

Follows the AgentSkills spec, so the same folder runs unchanged across OpenClaw, Claude Code, and other AgentSkills-compatible runtimes. Frontmatter keys are single-line, as required by OpenClaw's parser.

Structure

avatar-content-engine/
├── SKILL.md      # the skill (frontmatter + instructions)
└── README.md     # this file

License

MIT — see LICENSE.