Roles + conventions
Lead/dev/test/marketing (and more) so every agent has a clear job and lane.
How It Works
Most agent setups stop at: spawn a bot, write prompts, hope it remembers what you meant. That's not a team. That's chaos. ClawRecipes gives your OpenClaw setup real structure, and ClawKitchen makes it usable day-to-day.
ClawRecipes
Lead/dev/test/marketing (and more) so every agent has a clear job and lane.
Team memory and files everyone can read out of the box.
Work stays in your workspace: AGENTS.md, SOUL.md, tickets, outputs, and artifacts.
Intake → assign → execute → verify so teams can run without constant babysitting.
Add orchestrator/swarm patterns when you want multiple agents pushing one outcome in parallel.
ClawKitchen
Flow
Add ClawRecipes and ClawKitchen to OpenClaw so your system has both team scaffolding and a practical UI.
Start from a proven structure with agent roles, lane conventions, and operating defaults.
Edit team members, instructions, files, and workflows to match your real org and delivery process.
Run intake, assignment, execution, and verification loops with visible artifacts and file-based memory.
Install templates from the public marketplace and add parallel swarm/orchestrator patterns when needed.
Command Playbook
Install, verify, scaffold, and maintain your setup with repeatable commands.
Marketplace
There's a public marketplace angle too: install team templates + workflows that already work, then tailor them for your org.
Get Started
Built and maintained by @rjdjohnston.
Newsletter
Monthly updates on new recipes, workflows, and agent operations best practices.