Install rpiv-pi

Install rpiv-pi, wire its siblings, and run your first pipeline session.

rpiv-pi is the umbrella package. It ships the skill library and runtime infrastructure, but the tools the skills call live in sibling Pi extensions (rpiv-todo, rpiv-ask, rpiv-web, and friends). That’s why setup is three steps, not one.

1. Install the umbrella

From your shell:

pi install npm:@juicesharp/rpiv-pi

The npm: prefix tells Pi to resolve the package from the npm registry. Without it, Pi looks for a local path and the install fails.

2. Wire the siblings

Start a Pi session in any project, then run:

/rpiv-setup

/rpiv-setup reads ~/.pi/agent/settings.json, previews every missing sibling plugin alongside any legacy entries to prune, and waits for one confirm before applying. You see exactly what’s about to change.

3. Restart Pi

Quit and relaunch your Pi Agent session so the freshly-installed siblings load. The skills won’t find their tools until the runtime sees the new extensions.

Quick start: your first skill

Inside Pi, start the pipeline:

/skill:discover

The discover skill interviews you one question at a time to capture intent before any code is read. From there, rpiv-pi chains discover → research → blueprint → implement → validate. You keep answering questions, the skills do the work.

Next steps

• Pick your path: three workflows mapped to feature scope (small / mid / large)
• Walk the chain: the mid-size path demonstrated on a real example
• Reset between skills: the fresh-context rule between every transition
• Handoffs: pause the pipeline and resume from a fresh session
• Onboard a project: annotate a brownfield codebase before the first run
• Skills reference: every skill at a glance
• Agents reference: every specialist subagent at a glance