How to Leverage pm-product-discovery for Better Feature Development: A Complete Guide

The pm-product-discovery plugin provides a structured discovery engine with reusable skills and chained commands that transform vague ideas into validated, prioritized feature sets using proven product management frameworks.

The pm-product-discovery module in the phuryn/pm-skills repository offers a self-contained toolkit for guiding product teams through systematic user research and feature validation. By combining modular skills (domain-knowledge modules) with executable commands (end-to-end workflows), developers can implement data-driven discovery processes directly within Claude Code or any AI assistant that reads the markdown-based skill files.

Understanding the Architecture: Skills vs. Commands

The pm-product-discovery system organizes functionality into two distinct layers that work together to automate product discovery.

Skills: Reusable Domain Knowledge Modules

Skills are self-contained markdown files that encode proven PM frameworks as reusable building blocks. Each skill accepts specific inputs and returns structured outputs, making them composable across different discovery contexts.

Key skills include:

• brainstorm-ideas-new and brainstorm-ideas-existing – Generate feature concepts from multiple stakeholder perspectives, differentiating between greenfield products and existing platforms.
• identify-assumptions-new – Surfaces risky assumptions across six categories: Value, Usability, Viability, Feasibility, GTM (Go-to-Market), and Team.
• prioritize-assumptions – Ranks assumptions on an Impact × Risk matrix and suggests specific experiments to validate them.
• prioritize-features – Scores backlog items on Impact, Effort, Risk, and Strategic Alignment, outputting a ranked top-five list with rationale.
• opportunity-solution-tree – Implements Teresa Torres' OST framework, mapping outcomes to opportunities to solutions to experiments in a hierarchical structure.
• interview-script and summarize-interview – Create JTBD-focused interview guides and parse transcripts into structured insights.
• metrics-dashboard – Designs product metrics dashboards defining North-Star metrics, input metrics, and alert thresholds.

Source files for these skills reside in pm-product-discovery/skills/<skill-name>/SKILL.md, such as pm-product-discovery/skills/prioritize-features/SKILL.md which contains the core scoring logic.

Commands: Chained Workflow Execution

Commands orchestrate multiple skills into end-to-end workflows, handling user input, file I/O, and output formatting. They provide the CLI-style interface that developers interact with directly.

Core commands include:

• /discover – Runs the complete discovery loop: brainstorming, assumption identification, prioritization, experiment generation, and Opportunity Solution Tree creation.
• /brainstorm – Executes focused ideation sessions for new or existing products.
• /interview – Supports prep mode (generates scripts) or summarize mode (extracts insights from transcripts).
• /triage-requests – Categorizes and ranks batch feature requests against strategic criteria.
• /setup-metrics – Builds a complete metrics dashboard from a product vision statement.

Command definitions are stored in pm-product-discovery/commands/<command-name>.md, such as pm-product-discovery/commands/interview.md which wires together the interview-script and summarize-interview skills.

The Six-Step Discovery Workflow

Developers can leverage pm-product-discovery by chaining skills and commands through a systematic validation pipeline.

Step 1: Generate Ideas with Brainstorming

Initiate the process using the brainstorm-ideas-new skill or the /brainstorm command to collect raw feature concepts from multiple stakeholder lenses. This ensures ideation covers technical feasibility, user desirability, and business viability simultaneously.

Step 2: Surface Risk Through Assumption Mapping

Run identify-assumptions-new to expose hidden risks before committing development resources. The skill returns a structured table categorizing assumptions across Value, Viability, Feasibility, Usability, GTM, and Team dimensions.

Step 3: Prioritize by Impact and Risk

Feed the assumption table into prioritize-assumptions to score items on an Impact × Risk matrix. For concrete features, use prioritize-features (defined in pm-product-discovery/skills/prioritize-features/SKILL.md) which evaluates candidates across four criteria: Impact, Effort, Risk, and Strategic Alignment, emitting a ranked top-five list with detailed rationale.

Step 4: Map Opportunities with OST

For high-scoring opportunities, invoke opportunity-solution-tree to visualize the path from desired outcomes to specific solutions and experiments. This creates a hierarchical markdown artifact that serves as a living discovery document, directly implementing the framework found in pm-product-discovery/skills/opportunity-solution-tree/SKILL.md.

Step 5: Validate Through Customer Interviews

Use /interview prep to generate a JTBD-focused interview script based on the opportunity tree. After conducting interviews, call /interview summarize to extract Jobs-to-be-Done, pain points, satisfaction signals, and validated or invalidated assumptions. The output feeds directly back into the OST or feature prioritization table.

Step 6: Define Success Metrics

Finally, execute /setup-metrics or use the metrics-dashboard skill to establish quantitative validation criteria. The skill produces a markdown dashboard with a North-Star metric (e.g., "Summaries per active meeting"), supporting input metrics, target thresholds, and alert definitions to guide future iteration.

Practical Implementation Examples

Developers can invoke pm-product-discovery capabilities directly through Claude-style commands without writing custom code.

Running a Full Discovery Cycle

Execute the complete discovery loop with a single command:

/discover Build a new AI-powered meeting summarizer for remote teams

This automatically chains brainstorm-ideas-new, identify-assumptions-new, prioritize-assumptions, and brainstorm-experiments-new, outputting an Opportunity Solution Tree and prioritized feature list.

Prioritizing an Existing Backlog

Score specific features using the prioritize-features skill:

/prioritize-features
Feature ideas:
1. Real-time transcription
2. Shared meeting notes with tagging
3. AI-generated action items
4. Multi-language support
5. Export to project-management tools
6. Live sentiment overlay

The system returns a markdown table with individual scores for Impact, Effort, Risk, and Strategic Alignment, plus a final ranked list.

Preparing Customer Research

Generate interview scripts using the command defined in pm-product-discovery/commands/interview.md:

/interview prep Enterprise procurement workflow

The command prompts for research goals, participant details, and time constraints, then builds a script using the interview-script skill with a structured note-taking template.

Extracting Insights from Transcripts

Convert raw interview data into actionable intelligence:

/interview summarize
[Paste or upload the interview transcript here]

The summarize-interview skill parses the input into JTBD statements, current workflow descriptions, pain points, key quotes, and specific assumptions validated or invalidated during the conversation.

Building Metrics Dashboards

Define measurable success criteria:

/setup-metrics
Product: AI meeting summarizer
Goal: Reduce meeting-time waste by 30%

The metrics-dashboard skill generates a comprehensive dashboard including the North-Star metric, input metrics (adoption rate, summary accuracy), and alert conditions for monitoring implementation success.

Key Source Files for Developers

Understanding the underlying structure helps teams customize and extend the discovery toolkit:

File Path	Purpose
pm-product-discovery/skills/prioritize-features/SKILL.md	Core scoring algorithm for Impact/Effort/Risk/Strategic Alignment
pm-product-discovery/skills/identify-assumptions-new/SKILL.md	Risk matrix generation across six assumption categories
pm-product-discovery/skills/opportunity-solution-tree/SKILL.md	Teresa Torres OST implementation and hierarchy mapping
pm-product-discovery/skills/interview-script/SKILL.md	JTBD interview template generation
pm-product-discovery/skills/summarize-interview/SKILL.md	Transcript parsing and insight extraction logic
pm-product-discovery/skills/metrics-dashboard/SKILL.md	North-Star and input metrics framework
pm-product-discovery/commands/interview.md	Command wiring for interview prep and summarization
pm-product-discovery/commands/triage-requests.md	Batch feature request analysis workflow
pm-product-discovery/commands/setup-metrics.md	End-to-end metrics dashboard generation
README.md	Cross-plugin relationships and installation instructions

Summary

• pm-product-discovery provides a modular discovery engine through the phuryn/pm-skills repository, separating domain knowledge (skills) from workflow orchestration (commands).
• The system guides teams from ideation to validation through six structured steps: brainstorming, assumption mapping, prioritization, opportunity mapping, customer interviews, and metrics definition.
• Skills like prioritize-features and opportunity-solution-tree encode proven PM frameworks into reusable markdown modules located in pm-product-discovery/skills/.
• Commands such as /discover and /interview chain multiple skills into executable workflows defined in pm-product-discovery/commands/.
• Developers implement discovery workflows by calling commands directly in AI assistants, requiring no custom code while producing data-driven, validated feature roadmaps.

Frequently Asked Questions

How does pm-product-discovery differ from traditional product management tools?

Unlike static documentation or project management software, pm-product-discovery provides executable skills and commands that actively guide the discovery process. According to the source code in phuryn/pm-skills, each skill contains structured logic for specific PM frameworks (like Impact × Risk matrices or Opportunity Solution Trees), while commands automate the workflow chaining. This allows AI assistants to execute discovery steps interactively rather than merely storing the results.

Can I use individual skills without running full commands?

Yes. Developers can invoke individual skills directly by referencing the specific skill files. For example, you can call prioritize-features independently by providing a feature list directly to the skill defined in pm-product-discovery/skills/prioritize-features/SKILL.md. This modularity allows teams to integrate specific discovery activities (like assumption mapping or interview summarization) into existing workflows without adopting the entire command structure.

What inputs does the prioritize-features skill require?

The prioritize-features skill accepts a list of feature ideas and evaluates them against four criteria: Impact, Effort, Risk, and Strategic Alignment. As implemented in pm-product-discovery/skills/prioritize-features/SKILL.md, the skill outputs a markdown table with individual scores for each criterion and a final ranked list of the top five features, complete with rationale for each ranking decision.

How does the interview command handle transcript analysis?

The /interview command leverages the summarize-interview skill located at pm-product-discovery/skills/summarize-interview/SKILL.md to parse transcripts into structured outputs. The skill extracts Jobs-to-be-Done (JTBD), current workflow descriptions, pain points, key quotes, and specifically flags which assumptions were validated or invalidated during the conversation. This structured output feeds directly back into the Opportunity Solution Tree or feature prioritization tables.

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