How to Map Risky Assumptions for New Products Using PM Skills Marketplace

PM Skills Marketplace provides an AI-driven workflow that identifies, evaluates, and prioritizes risky assumptions across eight risk categories using the identify-assumptions-new and prioritize-assumptions skills.

The phuryn/pm-skills repository offers a systematic approach to map risky assumptions for new products through a markdown-based skill system. By leveraging the PM Skills Marketplace plugin, product managers can execute structured discovery workflows that surface hidden risks before committing engineering resources. The entire process runs within Claude-compatible environments using pure markdown instruction files that specify how AI models should analyze and categorize assumptions.

The Two-Skill Framework for Risk Mapping

The assumption mapping capability in PM Skills Marketplace centers on two discrete skills defined in the pm-product-discovery/skills/ directory. These skills operate sequentially to first surface potential risks and then evaluate them against quantitative criteria.

Identify Assumptions with identify-assumptions-new

The identify-assumptions-new skill, defined in pm-product-discovery/skills/identify-assumptions-new/SKILL.md, generates a structured list of assumptions across eight risk categories. When invoked, the skill prompts the AI model to analyze a product concept from three distinct perspectives: Product Manager, Designer, and Engineer.

The eight risk categories covered include:

• Value: Will users actually want this?
• Usability: Can users figure out how to use it?
• Viability: Can we monetize and sustain this?
• Feasibility: Can we build it with available technology?
• Ethics: Are there harmful unintended consequences?
• Go-to-Market: Can we reach and acquire users?
• Strategy & Objectives: Does this align with company goals?
• Team: Do we have the right talent and bandwidth?

Each assumption output includes a confidence rating, enabling downstream calculations.

Prioritize Assumptions with prioritize-assumptions

The prioritize-assumptions skill, located at pm-product-discovery/skills/prioritize-assumptions/SKILL.md, transforms the raw assumption list into an actionable matrix. This skill implements specific formulas referenced in the Domain Context section of the skill definition:

• Impact = Opportunity Score × Number of Customers
• Risk = 1 – (Confidence × Effort)

The skill places each assumption on an Impact × Risk matrix, categorizing items as high-impact/high-risk, high-impact/low-risk, low-impact/high-risk, or low-impact/low-risk. For every high-impact/high-risk item, the skill suggests a concrete experiment designed to validate or invalidate the assumption.

Executing the Complete Discovery Workflow

While individual skills handle specific tasks, the /discover command chains them into a comprehensive discovery process. The command definition in pm-product-discovery/commands/discover.md specifies the following execution sequence:

1. brainstorm-ideas-new
2. identify-assumptions-new
3. prioritize-assumptions
4. brainstorm-experiments-new

When you invoke /discover, the Claude runtime reads the command file, loads each referenced skill in order, and executes them as a single conversational flow. The command also auto-suggests next actions after each step, maintaining momentum through the discovery process.

Architecture and Implementation Details

The PM Skills Marketplace implements assumption mapping through a declarative architecture that requires no custom code execution. The system relies on three core components:

Skill Definition Files (*.md under pm-product-discovery/skills/) follow a uniform schema containing YAML front-matter, description, and step-by-step instructions. The AI model reads these instructions and executes the logic through natural language processing rather than programmatic function calls.

Command Files (*.md under pm-product-discovery/commands/) simply list ordered skill names. The Claude runtime interprets these lists as execution pipelines.

Plugin Manifest (pm-product-discovery/.claude-plugin/plugin.json) registers the plugin with the marketplace, exposing commands and skills to any Claude-compatible assistant. This manifest enables installation via the Claude Code marketplace, making the discovery toolkit available across different AI environments.

Because the skills are pure markdown instructions, they are language-agnostic and can be reused by any AI assistant that understands the PM Skills format, including Claude, Gemini, and OpenCode.

Practical Code Examples

Below are three practical ways to invoke the assumption-mapping workflow from a Claude-compatible environment.

Direct Skill Invocation

To generate an initial assumption list for a new product concept:

What are the risky assumptions for a new AI-writing-assistant that helps non-native speakers improve their essays?

Claude loads identify-assumptions-new automatically and replies with a markdown table of assumptions, each tagged with a confidence rating and risk category.

Prioritization Commands

After generating assumptions, trigger the prioritization logic:

Prioritize the assumptions you just listed.

Claude runs prioritize-assumptions, computes Impact × Risk values, and returns a categorized matrix. The output includes a table similar to:

Assumption	Impact	Risk	Category	Suggested Experiment
Users will pay $10/mo for premium feedback	High	Low	High Impact / Low Risk	Launch MVP with pricing test

End-to-End Discovery

To run the complete workflow in a single command:

/discover AI-writing-assistant for non-native English speakers

This executes the full chain: brainstorming ideas → identifying assumptions → prioritizing them → sketching experiments. The final output produces a concise discovery deck containing the product hypothesis, assumption matrix with experiments, and a next-step experiment plan.

Summary

• PM Skills Marketplace provides a markdown-based framework for mapping risky assumptions through the phuryn/pm-skills repository.

• The identify-assumptions-new skill surfaces assumptions across eight risk categories from PM, Designer, and Engineer perspectives.

• The prioritize-assumptions skill calculates Impact × Risk using the formulas Impact = Opportunity Score × # Customers and Risk = 1 – (Confidence × Effort).

• The /discover command chains skills into an end-to-end workflow defined in pm-product-discovery/commands/discover.md.

• All logic resides in markdown skill files under pm-product-discovery/skills/, making the system portable across AI assistants.

Frequently Asked Questions

What are the eight risk categories used in assumption mapping?

The eight categories are Value, Usability, Viability, Feasibility, Ethics, Go-to-Market, Strategy & Objectives, and Team. These categories ensure comprehensive coverage of product risks spanning user desirability, technical implementation, business sustainability, and organizational alignment.

How does the Impact × Risk matrix calculation work?

The calculation uses two formulas defined in the prioritize-assumptions skill Domain Context. Impact equals the Opportunity Score multiplied by the Number of Customers. Risk equals 1 minus the product of Confidence and Effort. These values position each assumption on a matrix that guides experiment prioritization.

Can I use PM Skills Marketplace without Claude?

Yes. Because the skills are pure markdown instructions without custom code dependencies, any AI assistant that can read and execute structured markdown can run the workflow. The plugin manifest in pm-product-discovery/.claude-plugin/plugin.json follows a standard format compatible with multiple AI marketplaces.

What is the difference between direct skill calls and the /discover command?

Direct skill calls invoke a single capability—such as identify-assumptions-new—in isolation. The /discover command chains multiple skills (brainstorm-ideas-new → identify-assumptions-new → prioritize-assumptions → brainstorm-experiments-new) into a guided conversational flow, automatically suggesting next steps and maintaining context between stages.

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