OpenClaw + lowkey viral

Let your agent make the videos

OpenClaw is an open-source AI agent framework. Agents built with it can discover tools through ClawHub, a skill marketplace. The lowkey viral skill teaches your agent how to call every endpoint in the lowkey viral API, so it can go from a topic to a finished video without you touching anything.

The skill is published under the slug lowkey-viral. Install it, hand your agent an API key, and it handles the full pipeline on its own: generating briefs, creating images, and rendering video.

Setup

Install the skill from ClawHub. Then create an API key in the API Keys dashboard. You need a PRO or ULTIMATE plan for API access (details on the FAQ page).

Your agent passes the key as a Bearer token:

Authorization: Bearer lkv_sk_your_key_here

That's it. The skill file contains the full API reference, so your agent knows which endpoints exist and what parameters they accept.

What the agent does

Here's the actual API sequence your agent will run. Each step is a standard HTTP call.

# Generate 5 content briefs from a prompt (2 credits)
curl -X POST https://lowkeyviral.com/api/v1/briefs \
  -H "Authorization: Bearer lkv_sk_your_key" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "We are Portland Roast Co., a specialty coffee roaster in Portland. We source single-origin beans and roast in small batches. Each post features 4 of our current offerings."}'

# Pick a brief and generate its images (4-8 credits)
curl -X POST https://lowkeyviral.com/api/v1/briefs/BRIEF_ID/generate \
  -H "Authorization: Bearer lkv_sk_your_key" \
  -H "Content-Type: application/json" \
  -d '{"image_model": "z_image_turbo"}'

# Start the video render (1 credit)
curl -X POST https://lowkeyviral.com/api/v1/briefs/BRIEF_ID/render \
  -H "Authorization: Bearer lkv_sk_your_key" \
  -H "Content-Type: application/json" \
  -d '{"design": "withCaptions", "soundtrack": "City Sunshine"}'

# Poll every 2-3 seconds until "type": "done"
curl https://lowkeyviral.com/api/v1/briefs/BRIEF_ID/render/RENDER_ID/progress \
  -H "Authorization: Bearer lkv_sk_your_key"

The result is a 1080x1920 MP4 at 30fps, five seconds long. You get an S3 URL back.

If you want fewer steps, use POST /api/v1/briefs/manual with "render": true. That collapses the entire pipeline into one call. The agent source on GitHub has the full parameter list.

Output formats

Your agent can produce two types of content:

Grid videos are 2x2 image layouts with text overlays, rendered as MP4. Three design options: default, withCaptions, and noSpaces.

Photo slideshows generate 4-10 individual JPEG slides with text. Caption styles are classic_bold, background_bar, and neon_glow.

For images, pick from three models. z_image_turbo costs 1 credit per image and is the fastest. p_image and flux_2_dev both cost 2 credits per image, with flux_2_dev producing more detailed results.

Practical uses

The real value here is batch automation. Give your agent a list of 20 prompts (each describing the brand and the topic for that post) and walk away. It generates briefs, picks image models, renders videos, and hands you download links. PRO accounts get 10 requests per minute, ULTIMATE gets 30.

This works well for agencies running multiple brand accounts. Set up a cron job or a scheduled task that feeds your OpenClaw agent fresh prompts daily, each one including the brand context so the briefs match the account's voice. Each run produces finished vertical videos ready to upload to TikTok or Instagram Reels.

You can also build approval workflows. Have the agent generate briefs, pause for human review, then continue with image generation and rendering only after sign-off. The multi-step API makes this straightforward because each stage is a separate call.

Create your API key and install the lowkey viral skill on ClawHub to get started. Full endpoint reference is in the API docs.