Showcase · The Big Four
These are the verbs the rest of the lexicon points to. Each one replaces a week of human glue work. Composed with the pipe operator, they ship a feature from "idea" to "edge-deployed" without you switching contexts once.
Sample terminal output below is illustrative. Real output varies per project and codebase.
Spawn the army.
Seven agents. One mission. Zero context-switching.
ll-swarm spawns a coordinated multi-agent fleet — coder, reviewer, tester, security, perf, design, deployer — in parallel git worktrees. They race; you keep the best. No sequential bottleneck. No single point of bluff.
npm install -g luna-agents · then /luna-agents:ll-swarm in Claude Code.
Apple polish, on every screen.
Audit, fix, ship. Like Cupertino, but in your terminal.
ll-hig reads your UI through Apple's Human Interface Guidelines — clarity, deference, depth, contrast, motion, accessibility — flags every violation, and writes the fix. SwiftUI-grade polish from a single command.
npm install -g luna-agents · then /luna-agents:ll-hig in Claude Code.
The app that fixes itself.
Test, screenshot, detect, fix. On loop. Until healthy.
ll-heal runs your suite, screenshots every route, detects regressions, writes the patch, and re-runs — until the app is green. Walk away. Come back to a healthy build.
npm install -g luna-agents · then /luna-agents:ll-heal in Claude Code.
One command. Everything fixed.
Code. Tests. Security. A11y. Perf. Deploy. Done.
ll-zen is the panic button. It runs the full restoration suite — readme-sync, no-bluf, hig, heal, security, perf, a11y, deploy — in dependency order, and stops only when every gate is green.
npm install -g luna-agents · then /luna-agents:ll-zen in Claude Code.
Star lunaos-ai/luna-agents on GitHub. Install in 30 seconds. Compose your first pipe in 2 minutes. Watch the swarm do a week of work in an afternoon.