Complete Guide to pm-product-discovery Plugin Commands: 5 Essential Slash Commands for Product Discovery

The pm-product-discovery plugin exposes five slash commands—brainstorm, discover, interview, setup-metrics, and triage-requests—that orchestrate thirteen underlying skills to automate end-to-end product discovery workflows within Claude.

The pm-product-discovery plugin in the phuryn/pm-skills repository transforms Claude into a specialized product management assistant. Designed for product managers running discovery workflows, this plugin maps natural language inputs to structured product methodologies through five primary pm-product-discovery plugin commands registered in the .claude-plugin/plugin.json manifest. Each command acts as a high-level orchestrator, triggering specific combinations of reusable skills stored in the pm-product-discovery/skills/ directory.

Architecture of the pm-product-discovery Plugin

The plugin follows a three-tier architecture that separates interface definitions from execution logic. Understanding this structure explains how the pm-product-discovery plugin commands translate user inputs into structured product outputs.

Plugin Manifest

The entry point resides at .claude-plugin/plugin.json, which declares the plugin name and registers the available slash commands with Claude's command system. This manifest file informs Claude which command files to expose to users.

Command Layer

Command definitions live in pm-product-discovery/commands/ as individual markdown files (e.g., brainstorm.md, discover.md). Each file documents the expected input schema and maps the command to one or more underlying skills. These files serve as the contract between the user interface and the execution layer.

Skills Layer

The actual logic resides in pm-product-discovery/skills/, containing thirteen distinct capabilities. Each skill occupies its own subdirectory with a SKILL.md specification file detailing inputs, outputs, and behavior. Skills are atomic and reusable; multiple commands can invoke the same skill.

The 5 pm-product-discovery Plugin Commands

The plugin exposes five slash commands that cover the full product discovery lifecycle, from ideation through experimentation design.

/pm-product-discovery:brainstorm

Purpose: Generates product ideas or lean experiments from multiple stakeholder perspectives.

Underlying Skills: This command orchestrates four skills: brainstorm-ideas-existing, brainstorm-ideas-new, brainstorm-experiments-existing, and brainstorm-experiments-new.

Functionality: Depending on parameters, it produces 5–10 ideas or experiment sketches tagged by perspective (Product Manager, Designer, or Engineer). For existing products, it builds upon current contexts; for new products, it generates early-stage concepts and pretotypes.

Source: Command definition located at pm-product-discovery/commands/brainstorm.md.

/pm-product-discovery:discover

Purpose: Executes a complete discovery cycle in a single invocation.

Underlying Skills: Combines brainstorming capabilities with identify-assumptions-existing (or identify-assumptions-new), opportunity-solution-tree, and prioritization skills.

Functionality: This command chains multiple skills sequentially: it first generates ideas, then identifies risky assumptions across Value-Usability-Viability-Feasibility quadrants, constructs an Opportunity Solution Tree (OST), and suggests experiments for high-impact assumptions.

Source: Implementation details in pm-product-discovery/commands/discover.md.

/pm-product-discovery:interview

Purpose: Supports customer research through script generation or transcript analysis.

Underlying Skills: Utilizes interview-script and summarize-interview.

Functionality: When invoked with action: "script", it outputs a structured interview protocol with JTBD (Jobs-to-be-Done) probing, warm-up questions, and wrap-up sections. When invoked with action: "summarise", it processes a transcript into a structured template highlighting satisfaction signals and action items.

Source: Specification found in pm-product-discovery/commands/interview.md.

/pm-product-discovery:setup-metrics

Purpose: Designs comprehensive product metrics dashboards.

Underlying Skills: Leverages the metrics-dashboard skill.

Functionality: Creates a measurement framework including North Star metrics, input metrics, health metrics, and alert thresholds. The output includes data source specifications, visualization recommendations, and threshold values for automated alerting.

Source: Documentation at pm-product-discovery/commands/setup-metrics.md.

/pm-product-discovery:triage-requests

Purpose: Analyzes and prioritizes batches of feature requests.

Underlying Skills: Powered by the analyze-feature-requests skill.

Functionality: Processes unstructured request lists, categorizes them by theme, evaluates strategic alignment, and scores each item by impact, effort, and risk. Returns a prioritized table with recommended next steps.

Source: Defined in pm-product-discovery/commands/triage-requests.md.

The 13 Underlying Skills

While users interact with the five pm-product-discovery plugin commands, the actual work is performed by thirteen specialized skills:

1. analyze-feature-requests – Prioritization by theme, alignment, impact, effort, and risk
2. brainstorm-experiments-existing – Experiment design for established products
3. brainstorm-experiments-new – Pretotype design for new ventures
4. brainstorm-ideas-existing – Multi-perspective ideation for current products
5. brainstorm-ideas-new – Early-phase feature generation
6. identify-assumptions-existing – Risk mapping across Value-Usability-Viability-Feasibility quadrants
7. identify-assumptions-new – Eight-category risk assessment including Go-to-Market and Strategy
8. interview-script – Structured customer interview creation
9. metrics-dashboard – North Star and input metric framework design
10. opportunity-solution-tree – OST construction mapping outcomes to experiments
11. prioritize-assumptions – Impact × Risk matrix analysis
12. prioritize-features – Backlog scoring by impact, effort, risk, and alignment
13. summarize-interview – Transcript synthesis into actionable JTBD insights

Each skill specification resides in pm-product-discovery/skills/<skill-name>/SKILL.md.

Code Examples: Invoking pm-product-discovery Plugin Commands

Below are the JSON payloads Claude sends when invoking each command. These examples demonstrate the input schemas defined in the command files.

Brainstorm Ideas or Experiments

{
  "command": "/pm-product-discovery:brainstorm",
  "input": {
    "product": "TaskTracker",
    "mode": "existing",
    "focus": "ideas"
  }
}

Result: Returns 5–10 ideas tagged by PM, Designer, and Engineer perspectives.

Run a Full Discovery Cycle

{
  "command": "/pm-product-discovery:discover",
  "input": {
    "product": "TaskTracker",
    "stage": "full"
  }
}

Result: Executes ideation → assumption identification → Opportunity Solution Tree construction → experiment suggestion.

Create or Summarize an Interview

{
  "command": "/pm-product-discovery:interview",
  "input": {
    "action": "script",
    "topic": "User onboarding"
  }
}

Result: A ready-to-use interview script with JTBD probing sections. Use action: "summarise" with a transcript to receive a structured summary instead.

Set Up a Metrics Dashboard

{
  "command": "/pm-product-discovery:setup-metrics",
  "input": {
    "product": "TaskTracker",
    "northStar": "Monthly Active Users",
    "inputMetrics": ["Sign-ups", "Feature Usage"],
    "healthMetrics": ["Retention", "Error Rate"]
  }
}

Result: A markdown table with metric definitions, data sources, and alert thresholds.

Triage Feature Requests

{
  "command": "/pm-product-discovery:triage-requests",
  "input": {
    "requests": [
      "Add dark mode",
      "Integrate with Slack",
      "Export CSV reports"
    ],
    "criteria": ["impact", "effort", "risk"]
  }
}

Result: A prioritized table with themes, strategic alignment scores, and next-step recommendations.

Summary

• The pm-product-discovery plugin provides five slash commands (brainstorm, discover, interview, setup-metrics, triage-requests) that serve as the primary interface for product discovery workflows.
• Each command orchestrates one or more of the thirteen underlying skills located in pm-product-discovery/skills/, ensuring atomic, reusable logic.
• Command definitions reside in pm-product-discovery/commands/ as markdown files specifying input schemas and skill mappings.
• The plugin manifest at .claude-plugin/plugin.json registers these commands with Claude, making them available as /pm-product-discovery:<command> slash inputs.
• All skills follow a standardized documentation format in their respective SKILL.md files, detailing inputs, outputs, and behavioral specifications.

Frequently Asked Questions

How do I install the pm-product-discovery plugin in Claude?

The plugin resides in the phuryn/pm-skills repository. To activate it, ensure the repository is configured as a Claude project plugin. The .claude-plugin/plugin.json manifest automatically registers the five slash commands (brainstorm, discover, interview, setup-metrics, triage-requests) with Claude's command interface, making them available immediately upon loading.

What is the difference between a command and a skill in the pm-product-discovery plugin?

Commands are user-facing entry points accessed via slash notation (e.g., /pm-product-discovery:discover). They orchestrate workflows by calling one or more skills. Skills are atomic, reusable capabilities defined in pm-product-discovery/skills/<skill-name>/SKILL.md that perform specific tasks like generating interview scripts or analyzing feature requests. A single command can trigger multiple skills in sequence.

Can I use the interview command to analyze existing transcripts?

Yes. The /pm-product-discovery:interview command supports two modes. Set action: "script" to generate a new customer interview script, or set action: "summarise" and provide a transcript to invoke the summarize-interview skill. The latter outputs a structured analysis including JTBD insights, satisfaction signals, and concrete action items.

Which command should I use for prioritizing a backlog of feature ideas?

Use /pm-product-discovery:triage-requests to analyze and prioritize existing feature requests, or /pm-product-discovery:discover for a comprehensive prioritization that includes assumption mapping and experiment design. The triage-requests command specifically utilizes the analyze-feature-requests skill to score items by impact, effort, risk, and strategic alignment.

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