- Home
- Skills
- Business & Operations
- Product Thinking: Turn a Raw Idea Into a Build-Ready Plan
More screenshots
Works with the AI Agents you already use
Product Thinking: Turn a Raw Idea Into a Build-Ready Plan
**Product Thinking** walks an account manager, salesperson, PM, or vibe coder from *"I want to build a thing"* to a plan an engineer (or Claude) can build from.
Free
Product Thinking: Turn a Raw Idea Into a Build-Ready Plan
Example session with this skill installed
I have an idea. I want a tool that pings account managers in Slack when a customer's contract is 60 days from renewal, and I want it deployed so the whole team can use it. Can you help me plan it?
- Read your context and instructions
- Compiled the product thinking
Summary
Lane
self-serve. Tier:
spike. Depth:
A (no revenue number to defend yet). No hosted app needed: this runs on a schedule once a day.
How we got here (5 questions, one at a time)
| Question | Your answer | What it decided |
|---|---|---|
| Who uses it, and how often? | 6 account managers, once a day | No always-on app. "Deploy it" was an assumption, not a need. |
| What happens in the 10 minutes after the ping? | AM looks up usage, then drafts a renewal email | The destination is the draft email, not the ping. |
| If a message fires wrongly, can you undo it? | Slack post yes, customer email no | Slack is automated. Customer email needs AM approval. |
| Where does the renewal date live? | HubSpot deal property | One access row: HubSpot private-app token. |
| Who else would read this list? | Finance forecasts renewals | Ledger row with a named second consumer. |
Build Sheet
Current state: someone checks the renewals spreadsheet on Mondays. Reminders go out late or never. The AM learns about it when the customer emails.
Ideal-state steps
- Every weekday at 9am, read deals renewing in 60 days.
- For each, pull last 90 days of product usage.
- Draft a renewal email using usage highlights.
- Post draft to the AM in Slack.
- AM approves; only then the email is sent. (approval gate)
Access checklist
| System | Credential | Who grants it | Verified |
|---|---|---|---|
| HubSpot | Private-app token (deals: read) | HubSpot admin | No |
| Slack | Bot token (chat:write) | Workspace admin | No |
| Usage data | Read-only DB user | Data engineering lead | No |
Shared-asset ledger
| Asset | Consumes / produces | System of record or second consumer | Exists today? |
|---|---|---|---|
| Upcoming renewals list | Produces | Finance (forecast) | No |
| Deal renewal dates | Consumes | HubSpot | Yes |
To confirm: where usage data lives (data engineering lead).
Schedule: weekdays 9am. Babysit for 2 weeks, then unattended.
Kill switch: disable the scheduled job; no app to take down.
Next steps
- Get the three access rows verified. No building until all three say yes.
- Paste this Build Sheet into a fresh session and ask for one supervised run on today's data.
- Before it runs unattended, send Finance a three-line note about the renewals list.
Connects securely to your tools. The creator never sees your data.
About this skill
AI coding tools don't stop and ask why. The person with the idea usually already knows who it's for, how many people will use it, and whether the data needs to stick around. Nothing pulls those answers out, so the AI builds for a million users when you have five.
Product Thinking walks an account manager, salesperson, PM, or vibe coder from "I want to build a thing" to a plan an engineer (or Claude) can build from. You never face a "left or right" technical question. You answer plain questions about your work, and the skill makes the technical call for you and tells you why.
What it figures out
- Why and outcome: what it's for, checked back against what you're building.
- Demand check: what people do today, what it costs them, and one named person who wants it.
- Build, borrow, or buy: does something you already pay for do this? If so, you stop here with an adoption note instead of a build.
- Scope and right-sizing: who uses it, how often, what data, how fresh. Traffic is estimated bottom-up, with a scale ceiling so the build fits your real number of users.
- Production readiness: data sensitivity, who can view or edit, what it connects to, who owns it, and what's out of scope for v1.
- Shared-asset ledger: what the tool consumes and produces, so you don't quietly create another copy of your CRM.
Three gates
- Lane: self-serve automations you run yourself end with a one-page Build Sheet (ideal steps, approval gates for anything irreversible, access checklist, one supervised run, then a schedule). Everything else gets an engineering handoff.
- Tier: spike, app, or production. Sets how much build rigor it needs.
- Depth: a short Product Brief for most internal tools. The full Goal Brief (outcome, thesis, milestone, metrics) only when a real business number is on the line.
How it talks to you
It counter-argues and never vetoes. It backs off when your answer holds up. It names the hidden assumption ("deploy it", "it needs a database", "this drives revenue", "just notify someone") and tests it against reality. If your engineering team has written standards, it defers to them.
The proof
There is no measured result yet, and the listing claims none. The checks are concrete. A Why Gate comes before any scoping question. Traffic is estimated as units times adoption times frequency, and the plan carries a scale ceiling so the build fits your real number of users. Four common assumptions are named and tested on sight: that it must be deployed, that it needs a database, that it drives revenue, and that someone will be notified. Self-serve automations end with a Build Sheet that puts a human approval gate on every irreversible step and requires one supervised run before any schedule. The plugin passed validation and was install-tested from GitHub in a clean home directory.
Use it
Say "I have an idea for a tool that..." or "I want to automate this thing I do every week." In Claude Code it also installs as a plugin with four command aliases: /product-thinking:help-me-build, :help-me-think, :i-have-an-idea, :plan-my-idea.
Source and license
Free to use and MIT licensed. The full source, install steps, and issue tracker are on GitHub at https://github.com/robertmnyborg/product-thinking.
Plans only: it does not write the code or a full engineering spec. It hands off to your own build process. Expect a few passes; the first plan rarely connects cleanly to the outcome, and that's the method working.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 8 days ago
- Free with an Agensi account
Permissions
Reads files only to understand your project when you point it at one. Writes a single markdown file: the Product Brief, Goal Brief, or Build Sheet it produces. No network calls, no shell commands, no credentials.