Terminology

A quick overview of the pieces and how they fit together.


The Four Pieces

Skills

Slash commands that do one product management job. Type /prd-generator and Claude walks you through writing a PRD. Type /competitive-profile-builder and it analyzes a competitor.

Each skill follows a proven product management framework (Marty Cagan, Teresa Torres, April Dunford) and produces structured output you can use immediately.

70+ skills included — from discovery to strategy to communication.

Workflows

Multi-step processes that chain skills together. Instead of running one skill at a time, a workflow runs an entire sequence — sometimes with multiple Claude agents working in parallel.

Example: The Batch Interview Analysis workflow processes 10 interview transcripts simultaneously, then synthesizes findings across all of them. What takes 4 hours manually takes 20 minutes.

Templates

Output formats that skills use. A PRD always follows the same structure. A status update always hits the same sections. Templates give you consistency without you thinking about formatting.

You can customize templates in the templates/ folder if the defaults don't match your team's format.

Context

Your company, product, personas, and competitors — stored in markdown files in the context/ folder. Every skill reads these automatically.

This is the difference between Second and generic AI. You never re-explain your product. Every output references your actual users, competitors, and priorities — not generic placeholders.


How They Connect

Context (your company info)
    ↓ feeds into
Skills (slash commands)
    ↓ use
Templates (output formats)
    ↓ chain into
Workflows (multi-step processes)

Context makes everything personalized. Skills do the work. Templates keep output consistent. Workflows handle complex multi-step tasks.


The Inputs/Outputs Pattern

Every work folder follows the same structure:

discovery/
├── inputs/    ← Drop files here (transcripts, research)
└── outputs/   ← Second saves results here

strategy/
├── inputs/    ← Drop files here (competitor info)
└── outputs/   ← Second saves results here

You always know where to put things and where to find results.


Why Context Matters

Without ContextWith Context
"Write a PRD for a login feature" → generic template"Write a PRD for a login feature" → references your product, users, metrics
"Analyze Competitor X" → surface-level comparison"Analyze Competitor X" → positions against your actual strengths
"Write a status update" → fill-in-the-blank template"Write a status update" → pulls from your goals and current priorities

The more specific your context files, the better every output. You can enrich your context anytime.