Autonomous website improvement loop
An infinite optimization loop for websites. SiteLoop continuously measures metrics, analyzes opportunities, proposes changes, implements winners, evaluates impact, and repeats. It's like having a full growth team that never sleeps.
The loop runs through seven phases: Measure → Analyze → Propose → Challenge → Implement → Evaluate → Record. Each phase can be handled by different agent roles in a swarm configuration.
Websites don't improve themselves — or they didn't, until now. Most sites launch and stagnate. SiteLoop means continuous, data-driven improvement without manual intervention. Set it up, let it run, watch your metrics climb.
siteloop.json configuration schemaAfter launching oriclaw.com, I realized the hard part isn't building the site — it's improving it after launch. Most sites ship once and stagnate. I wanted something that would keep optimizing autonomously: measure what's working, propose changes, test them, keep what wins. A growth team that runs while I sleep. This is the protocol I designed for that — and it's what keeps my own site getting better.
# Configure your site
cp config.example.json siteloop.json
# Edit with your site details, metrics sources, deploy command
# The loop runs autonomously:
# MEASURE → ANALYZE → PROPOSE → CHALLENGE → IMPLEMENT → EVALUATE → RECORD
Unzip into ~/.openclaw/workspace/skills/ and read the SKILL.md inside.