Give Claude Code
an opinionated playbook
The playbook we run our own client work on: markdown skills, focused agents, and a plan → execute → verify loop for Claude Code. All plain files — inspectable, editable, deletable.
Phased Workflow
Brainstorm, plan, execute, and verify. A repeatable loop for the work that deserves structure — ad-hoc sessions stay ad-hoc.
Learn more →Stack Profiles
Pre-built skill bundles for Django, Next.js, Flutter, Astro, and a global base. One profile installs the conventions and verification checklists for your stack — no re-explaining it every session.
Learn more →Composable Skills
Markdown files that teach Claude patterns, conventions, and guardrails. Mix and match what you need — nothing is locked in.
On-Demand Commands
Slash commands like /scaffold, /prd, and /verify — each is just a skill with auto-triggering turned off, so it costs zero context until you type it.
Cross-Repo Coordination
Integration summaries and shared conventions that keep agents aligned across frontend, backend, and infrastructure repos.
Dedicated Agents
Persona-driven agents for planning, execution, and review. Each one scoped to a role with its own tools, tone, and guardrails.