Graphic Designer Agent

A Claw configured as a graphic designer produces visual assets on demand — slide decks, social graphics, brand materials, and video content — using a content pipeline that handles research, writing, AI video generation, and publishing in a single automated flow.

What this agent does

A graphic designer Claw handles the full content production loop:

ESSAY_SOUL.md — your voice file

Before this agent writes a single word, it needs to understand who it's writing for. ESSAY_SOUL.md is the author's voice file — it captures your tone and personality, the principles and practices you hold, and your core approach to everything you create. It's what makes the output sound like you and not like a generic AI script.

Create it in your workspace root and include:

# Essay Soul
 
## Voice and tone
[How you sound — e.g. direct, opinionated, anti-hype, conversational]
 
## Principles I hold
[What you believe about your subject — the lens you apply]
 
## Things I won't do
[What you avoid — e.g. listicles, hedging, motivational language]
 
## My approach
[How you structure arguments, what you find interesting, what bores you]

The Claw reads this before every writing task. It's the difference between a script that sounds like you gave a talk and one that sounds like a press release.

Setup

npx clawhub@latest install humanizer
npx clawhub@latest install agent-browser

openclaw env set HEYGEN_API_KEY=your_key
openclaw env set WRITER_API_KEY=your_key  # optional

Update ~/TOOLS.md to note:
- Humanizer skill installed
- HeyGen API key configured
- ESSAY_SOUL.md exists at workspace root
- YouTube OAuth credentials at client_secrets.json

Read ~/TOOLS.md and ESSAY_SOUL.md.
Research this URL: [link to something you want to cover]
Write a 280-word essay in my voice.
Show me the draft before we go further.

The content pipeline

Once configured, the full run from URL to uploaded video:

URL → Research → Essay → Proof → Humanize → HeyGen Video → Bumpers → YouTube

Each step is a separate Claw task. The agent saves output artifacts to content/research/, content/essays/, and content/videos/ so any step can be re-run independently.

See the Content Creation Pipeline recipe for the complete step-by-step walkthrough including HeyGen API configuration, bumper stitching, and upload scheduling.

HTML slide decks

The same agent can produce presentation decks from an outline. Define your brand guide once:

My brand: background #0D1B2A, accent #F4A623, font Inter.
Build a slides template with these slide types:
title, agenda, section, statement, bullets, steps, close.
Save to ~/templates/slides-template.html

Then generate any deck:

Read ~/templates/slides-template.html.
Build a 20-slide deck from this outline: [outline]
Publish to ZenBin as [slug].

Full walkthrough at zenbin.org/p/tutorial-build-html-slides (opens in a new tab).

Scheduling content production

Run the pipeline on a schedule with Clawvisor connected:

openclaw cron add \
  --name "Weekly content" \
  --cron "0 8 * * 1" \
  --session isolated \
  --message "Read ~/TOOLS.md and ESSAY_SOUL.md.
Check my content queue in Notion.
Pick the top item, run the full content pipeline,
upload to YouTube as unlisted.
Send me the YouTube URL via Telegram." \
  --announce \
  --channel telegram