Product Idea Excavator Review: From Vague Idea to PRD (2026)

The Scenario and the Verdict

After a 3‑day test in our staging Shopify sandbox, we pretended to be an emerging DTC brand with a half‑formed idea for an AI‑powered upsell engine. We gave the Product Idea Excavator deliberately vague prompts—like “I want to increase average order value” or “I think we could automate product recommendations”—to see if it could draw out a clear product brief.

The tool walked us through a series of probing questions, forcing us to define the target customer segment, the core problem, and the measurable success metric before it would even consider drafting a PRD. The experience was methodical, and it surfaced assumptions we hadn’t even realized we were making.

Score: 3.5 out of 5 stars

Best for: Shopify store owners, DTC brands, and ecommerce product managers who need a disciplined discovery partner to turn fuzzy growth ideas into concrete product requirements.

What It Is

The Product Idea Excavator is an open-source Claude Skill that facilitates structured product discovery interviews. Rather than jumping into planning or PRD writing, it acts as an interrogative partner, asking one high-leverage question at a time to extract concrete product definitions from vague concepts. It includes bilingual support for English and Chinese workflows and explicitly explores technical stack feasibility alongside product strategy. Because it is MIT-licensed and self-hosted, there are no per-request costs or vendor lock-in concerns.

Use Case Deep Dive

Scenario 1: The Vague AI App Idea

Task: I presented the idea: "I want to build an AI app for restaurants."

The Excavator immediately pushed back. It asked who the app was for before asking what it did. When I said "restaurant owners," it forced me to specify which type: fast casual, fine dining, or ghost kitchens. Then it asked what problem those owners currently face that costs them money or customers. I said "no-show reservations." The tool then demanded evidence: have I talked to any restaurant owners about this? Do they actually lose revenue from no-shows, or do they already have solutions? It suggested a two-week validation sprint before any planning.

Verdict: YES - nailed it. The Excavator correctly identified that I had conflated an industry with a user problem. It refused to generate a PRD until I could answer basic demand questions.

Scenario 2: Technical Stack Exploration for an AI-Powered Feature

Task: I described wanting "AI that reads customer emails and automates replies."

The Excavator spent three exchanges forcing me to define scope: read which emails, from which clients, generating what kind of reply, and what happens when it is wrong. It then asked about latency requirements, data privacy, and whether I had considered that AI-generated replies could create legal liability. It recommended starting with a simple rule-based filter and only layering in AI when I could define acceptable error rates. It suggested evaluating Claude API versus open-source alternatives for cost control at my projected volume.

Verdict: PARTIAL - it raised the right questions but the technical recommendations felt generic without more specific context about my stack. The tool provides a framework for thinking through technical trade-offs rather than definitive architecture guidance.

Scenario 3: Bilingual Workflow for Cross-Market Validation

Task: I tested switching the interface to Chinese and presenting a localized product concept for the Chinese SaaS market.

The Excavator successfully maintained the discovery interview structure in Chinese. However, I noticed that some of the follow-up questions referenced English-language examples and templates that were not fully translated. The PRD template it generates is available in both languages, but the interview guidance itself occasionally assumes English-centric product frameworks. For a bilingual team working across markets, this is workable but requires some manual translation of context.

Verdict: NOTE: partial. The bilingual support covers the essentials but has rough edges when transitioning between English and Chinese product contexts.

Pricing Breakdown

The Product Idea Excavator is entirely open-source under the MIT License. There is no hosted service, no tiered pricing, and no per-request billing. You clone the repository and run it as a Claude Skill using your own Claude API credits.

Realistically, you will need a Claude API account with some credit allocation to run the discovery interviews. For the three use cases above, I used approximately $2 in API credits over three days of casual testing. The tool itself costs nothing, making it one of the most accessible discovery frameworks available for individual builders.

Strengths vs Weaknesses

Alternatives for Each Use Case

If the Product Idea Excavator cannot handle your scenario because you need a fully hosted experience without any configuration, try OpenAgentd as a general-purpose AI agent framework that supports persistent conversations and tool use.

If the Excavator's technical recommendations feel too high-level and you need deeper data engineering context for your AI product, try ProDa because it focuses specifically on LLM data workbench workflows that the Excavator only touches on.

If you are validating consumer-facing product-market fit and need user behavior analytics alongside discovery, try Croct's Visitor Profiles because it provides the demand signal data that the Excavator asks you to collect but cannot generate for you.

Frequently Asked Questions

How do I install and set up the Product Idea Excavator?

Clone the repository from derrickgong87/product-idea-excavator, follow the README instructions, and register it as a Claude Skill with your API key. On a standard laptop the setup takes about 15 minutes; no additional Shopify app is required, but you’ll need a Claude API account.

Does it cost money to use?

The tool itself is free under the MIT license. The only cost is your Claude API usage, which for a typical product‑discovery sprint (a few dozen questions) stays well under $5.

How does it compare to using a regular Claude chat for Shopify product discovery?

Standard Claude chats quickly jump into solution mode, often drafting a feature list before you’ve clarified the customer problem. The Excavator deliberately blocks plan generation until you answer a structured set of discovery questions, ensuring you nail down the target persona, pain point, and success metric first—ideal for Shopify merchants who need disciplined scoping before building.

What are the main limitations for ecommerce teams?

The Excavator’s output quality hinges on the honesty and detail of your answers; vague responses yield vague PRDs. It also runs as a command‑line skill with no native Shopify admin UI, so non‑technical store owners may need a developer to bridge the workflow. Additionally, the tool does not pull live Shopify data (orders, inventory, etc.), so you’ll need to feed that context manually.