This handles the tedious part of starting or resuming a Claude project by creating a consistent folder structure (specs, scripts, notebooks, manifests, agent_memory) and resolving what to call everything. It checks for existing PLAN.md files first so you don't accidentally create duplicates, then either uses context from your conversation or asks you to confirm a slugified project name. The structure is opinionated but sensible, keeping generated code separate from plans and using agent_memory for session notes that stay out of git. It's the kind of thing you'd otherwise copy-paste from project to project or forget to do until you have files scattered everywhere.
npx -y skills add awslabs/agent-plugins --skill directory-management --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
metabase/metabase
github/awesome-copilot
UKGovernmentBEIS/inspect_evals
addyosmani/agent-skills