What Are the Four Skills Chained by the /discover Command in pm-skills?

The /discover command sequences four specialized skills—Brainstorm Ideas, Identify Assumptions, Prioritize Assumptions, and Brainstorm Experiments—to transform product concepts into validated experiments.

The phuryn/pm-skills repository provides structured product management workflows for AI-assisted discovery. The /discover command serves as the primary entry point, automatically sequencing four specialized skills to move from vague concepts to testable experiments. Understanding these four skills chained by the /discover command is essential for product managers leveraging this open-source framework.

The Four Skills in the Discovery Pipeline

The complete workflow is orchestrated through pm-product-discovery/commands/discover.md, which stitches together four distinct phases. Each phase corresponds to a specific skill, with separate implementations for existing products versus new products where context demands different approaches.

1. Brainstorm Ideas

This skill generates a diverse set of product ideas from Product Manager, Designer, and Engineer perspectives, presenting the top ten candidates for user selection. Implemented in brainstorm-ideas-existing.md and brainstorm-ideas-new.md, this skill feeds directly into the assumption identification phase.

2. Identify Assumptions

This skill surfaces critical assumptions across value, usability, feasibility, and viability dimensions—plus go-to-market considerations for new products. Using a multi-perspective "devil's advocate" analysis, the identify-assumptions-existing and identify-assumptions-new variants challenge each selected idea to expose hidden risks before prioritization.

3. Prioritize Assumptions

The prioritize-assumptions skill maps extracted assumptions onto an Impact × Risk matrix, isolating "leap-of-faith" assumptions that combine high impact with high uncertainty. Unlike the other skills, this single implementation serves both existing and new product workflows, ranking assumptions to determine which require immediate validation.

4. Brainstorm Experiments

This final skill designs 1–2 validation experiments per critical assumption. The brainstorm-experiments-existing variant suggests A/B tests, fake doors, and prototypes, while brainstorm-experiments-new recommends pretotypes, landing pages, and concierge MVPs. Each experiment targets specific leap-of-faith assumptions identified in the previous step.

Source File Structure

The skill chain is defined in pm-product-discovery/commands/discover.md, which references eight skill implementation files:

• Idea Generation: pm-product-discovery/skills/brainstorm-ideas-existing.md and brainstorm-ideas-new.md
• Assumption Extraction: pm-product-discovery/skills/identify-assumptions-existing.md and identify-assumptions-new.md
• Prioritization: pm-product-discovery/skills/prioritize-assumptions.md (shared across both product types)
• Experiment Design: pm-product-discovery/skills/brainstorm-experiments-existing.md and brainstorm-experiments-new.md

Usage Examples

Basic Invocation

Running the command with a product concept initiates the full four-skill chain:

/discover Smart notification system for our project management tool

This executes: Brainstorm Ideas → Identify Assumptions → Prioritize Assumptions → Brainstorm Experiments.

Interactive Mode

Invoking the command without arguments triggers interactive mode:

/discover

The system asks "What are you exploring?" and then proceeds through the same four-skill pipeline after receiving your input.

Selecting Ideas for Validation

After the brainstorming step, the system presents results and waits for selection:

Here are 10 ideas. Which ones should we stress-test? Pick 3-5, or I can carry all forward.

Your selection feeds directly into the Identify Assumptions skill, after which the chain continues automatically through prioritization and experiment design.

Summary

• The /discover command chains four sequential skills to automate product discovery workflows.
• Brainstorm Ideas generates multi-perspective product concepts using brainstorm-ideas-existing or brainstorm-ideas-new.
• Identify Assumptions surfaces risks via devil's-advocate analysis using identify-assumptions-existing or identify-assumptions-new.
• Prioritize Assumptions ranks assumptions using an Impact × Risk matrix via the shared prioritize-assumptions skill.
• Brainstorm Experiments designs validation tests using brainstorm-experiments-existing or brainstorm-experiments-new.

Frequently Asked Questions

What is the difference between the existing and new product variants?

The Brainstorm Ideas, Identify Assumptions, and Brainstorm Experiments skills maintain separate implementations for existing versus new products. Existing product variants focus on feature enhancements, A/B tests, and integration with current systems, while new product variants emphasize go-to-market assumptions, pretotypes, and concierge MVPs. The Prioritize Assumptions skill uses a single implementation for both contexts.

Can I run individual skills without the full /discover chain?

While the /discover command automatically sequences all four skills, the repository structure in phuryn/pm-skills exposes each skill as a standalone module. You can invoke brainstorm-ideas-existing, identify-assumptions-new, or other specific skills independently if you need to skip steps or repeat a specific phase of the discovery process.

What happens if I provide no initial idea to the /discover command?

Running /discover without arguments initiates interactive mode. The system prompts "What are you exploring?" and waits for your input before proceeding through the four-skill chain. This allows for conversational discovery without requiring a pre-formed hypothesis.

How does the prioritize-assumptions skill determine which assumptions to test first?

The prioritize-assumptions skill maps each assumption onto an Impact × Risk matrix. Assumptions landing in the high-impact, high-uncertainty quadrant—termed "leap-of-faith" assumptions—receive priority ranking. This ensures the subsequent experiment design phase targets the riskiest elements first, maximizing validation efficiency.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}