How to use the Identify New Product Assumptions agent skill

What is an agent skill and what does this one do?

An agent skill is a folder of instructions that an AI agent can load when a task matches it. This skill is defined in a single file called SKILL.md, which tells the agent how to approach a specific kind of work.

This skill covers new product ideas. The agent looks at the idea from three viewpoints (product manager, designer and engineer) and checks it against eight risk areas: value, usability, viability, feasibility, ethics, go-to-market, strategy and objectives, and team. For each assumption it finds, it rates how confident you should be and suggests a way to test it. The result is saved as a markdown document.

• The four core product risks (value, usability, viability, feasibility) come from Teresa Torres's Continuous Discovery Habits.
• The other four areas (ethics, go-to-market, strategy and objectives, team) are added for new products.
• The skill reasons only from your prompt and any files you supply. It does not browse or verify claims.

Install the identify-assumptions-new skill

Download the ZIP file for this skill from this page. Then install it in one of two ways.

You can unzip it into your personal skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill available only to that project.

Alternatively, you can upload the skill in the Claude apps. Open Settings, go to Capabilities, then choose Skills and upload the ZIP.

• Personal install: unzip into ~/.claude/skills/
• Project install: unzip into the project's .claude/skills/ folder
• Claude apps: Settings, then Capabilities, then Skills, then upload

How do you trigger the skill with a natural request?

You do not need a special command. The skill is meant to be picked up when your request matches its description, which covers evaluating startup risks, assessing a new product concept, or mapping assumptions for a new venture.

The skill works best when your request includes the product concept, the target segment and the feature idea. The skill's own instructions say it expects this, and that it will read any files you provide, such as business plans or research, before it starts.

• Assess the startup risks for my idea: a meal-planning app for busy parents.
• Map the assumptions for a new venture that sells refurbished e-bikes to university students.
• Here is my business plan. Identify the risky assumptions in this new product concept.
• Stress-test my idea for a B2B invoicing tool for freelancers, focusing on go-to-market and team risks.

What output should you expect?

The agent works through the eight categories and lists the assumptions it finds under each one. For every assumption, it gives a confidence rating and a suggested test. The skill asks for the output to be saved as markdown, so you get a document you can keep, edit or share with your team.

Treat the ratings as the agent's judgement about your idea, not as evidence. The skill does not check market data, so any claim it makes about demand or competitors should be verified separately.

• Assumptions grouped under the eight risk categories
• A confidence rating for each assumption
• A suggested test for each assumption
• A markdown file containing the full review

What does the skill need and what are the safety notes?

The listing lists no requirements for this skill. It does not run scripts and does not use the network, so it works only from the text of your request and any files you give it.

Because it does not browse, the skill cannot confirm market size, competitor details or regulatory status. It mentions compliance and legal factors, but it gives no professional advice. For legal, regulatory or financial decisions, consult a qualified lawyer or accountant.

• Requires: nothing listed
• Runs scripts: no
• Uses network: no
• Not a source of market data or verified facts
• Not professional legal or financial advice

Tips for getting useful results

The more specific your input, the more specific the assumptions. A vague idea produces vague risks. Name the customer, the problem and the first version of the product before you ask.

Use the output as a checklist for your next conversations. Each suggested test is a prompt for a real experiment, such as an interview, a landing page or a prototype, that you can run yourself.

• Give the target segment and the core feature idea in the same message.
• Attach business plans or research files so the agent reads them first.
• Pick the two or three assumptions with the lowest confidence and test those first.
• Re-run the skill after you learn something new, and compare the results.