How to Use the /discover Command in the pm-product-discovery Plugin

The /discover command initiates a structured product discovery workflow that chains together brainstorming, assumption identification, risk prioritization, and experiment design to validate product ideas.

The /discover command in the pm-product-discovery plugin automates a comprehensive four-stage discovery process. Found in the phuryn/pm-skills repository, this command helps product teams systematically explore ideas, surface risks, and design validation experiments. Whether you are conducting continuous discovery on an existing product or exploring a new concept, understanding how to use the /discover command streamlines your research workflow.

What Does the /discover Command Do?

The command executes a sequential workflow defined in pm-product-discovery/commands/discover.md that combines four distinct skills:

• Brainstorm ideas – Generates product concepts from PM, Designer, and Engineer perspectives using the skill defined in pm-product-discovery/skills/brainstorm-ideas-new/SKILL.md.
• Identify assumptions – Extracts risks and hypotheses from the generated ideas, implemented in pm-product-discovery/skills/identify-assumptions-new/SKILL.md.
• Prioritize assumptions – Ranks risks using the Opportunity Score (calculated as Importance × (1 − Satisfaction)), handled by pm-product-discovery/skills/prioritize-assumptions/SKILL.md.
• Brainstorm experiments – Designs validation experiments for top-ranked assumptions via pm-product-discovery/skills/brainstorm-experiments-existing/SKILL.md.

How to Run the /discover Command

Basic Syntax

You can invoke the command with or without arguments:

/discover Smart notification system for our project-management tool

If you omit the argument, the system prompts: "What are you discovering?"

/discover

Specifying the Product Stage

Explicitly declare whether you are exploring a new product or conducting continuous discovery on an existing one:

/discover New product: AI writing assistant for non-native speakers --stage new

The command declaration in .claude-plugin/plugin.json under the "commands" section registers this interface, while the detailed documentation in pm-product-discovery/commands/discover.md explains the stage parameter.

The Four-Step Discovery Workflow

When executed, /discover chains the following skills in sequence:

Step 1: Brainstorm Ideas

The workflow begins by generating diverse product concepts. The system solicits perspectives from product management, design, and engineering roles to ensure comprehensive coverage of potential solutions.

Step 2: Identify Assumptions

Next, the command extracts implicit and explicit risks from the brainstormed ideas. This step surfaces the hypotheses that must hold true for the product to succeed.

Step 3: Prioritize Assumptions

Risks are scored and ranked using the Opportunity Score formula: Importance × (1 − Satisfaction). Assumptions with high importance and low current satisfaction score highest, indicating they offer the greatest validation opportunity.

Step 4: Brainstorm Experiments

Finally, the system generates concrete experiment designs to validate the top-ranked assumptions, providing actionable next steps for your discovery process.

Understanding the Output

After completion, /discover presents a structured Discovery Plan containing:

• Brainstormed Ideas from multiple stakeholder perspectives
• Key Assumptions requiring validation
• Prioritised Risks ranked by Opportunity Score
• Experiment Designs with validation methods

You can treat this as a living document, requesting updates as experiments progress and new data emerges.

Key Implementation Files

The command relies on these specific files in the phuryn/pm-skills repository:

• .claude-plugin/plugin.json – Declares the plugin metadata and registers the /discover command in the "commands" section.
• pm-product-discovery/commands/discover.md – Contains user-facing documentation explaining syntax, workflow steps, and examples.
• pm-product-discovery/skills/brainstorm-ideas-new/SKILL.md – Implements the initial idea generation phase.
• pm-product-discovery/skills/identify-assumptions-new/SKILL.md – Handles risk extraction from generated concepts.
• pm-product-discovery/skills/prioritize-assumptions/SKILL.md – Calculates Opportunity Score and ranks assumptions.
• pm-product-discovery/skills/brainstorm-experiments-existing/SKILL.md – Produces validation experiments for prioritized risks.

Summary

• The /discover command chains four skills to automate product discovery workflows.
• Invocation supports optional product descriptions and --stage flags for new or existing products.
• The prioritization engine uses the Opportunity Score (Importance × (1 − Satisfaction)) to rank risks.
• Output includes a comprehensive Discovery Plan with ideas, assumptions, rankings, and experiments.
• Command registration occurs in .claude-plugin/plugin.json, with logic distributed across skill-specific files.

Frequently Asked Questions

What inputs does the /discover command require?

The command accepts an optional product description as an argument. If omitted, it prompts you with "What are you discovering?" It also asks whether you are exploring an existing product (continuous discovery) or a new product (initial discovery), and what decisions the discovery should inform, such as build/kill, prioritize, or pivot.

How does the prioritization scoring work?

The system ranks assumptions using the Opportunity Score formula: Importance × (1 − Satisfaction). This calculation, implemented in pm-product-discovery/skills/prioritize-assumptions/SKILL.md, identifies assumptions that are highly important but currently have low satisfaction, indicating high validation potential.

Can I customize the discovery workflow?

While the command follows a fixed four-step sequence defined in the source files, you can modify individual skill behaviors by editing the respective SKILL.md files in the pm-product-discovery/skills/ directory. The output serves as a living document that you can update and expand as your discovery process evolves.

Where is the /discover command defined in the codebase?

The command is declared in .claude-plugin/plugin.json under the "commands" section, which registers the slash command with the system. The detailed user documentation and workflow logic reside in pm-product-discovery/commands/discover.md, while the execution steps are implemented across the skill files in the pm-product-discovery/skills/ directory.

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