This is a submission for the GitHub Copilot CLI Challenge
What I Built
The Problem: The "Theory-to-Practice" Gap
As a software engineer trying to transition into Product Management, I hit a massive wall: The Experience Paradox.
- To get a PM job, you need a portfolio of case studies.
- To build a portfolio, you need real-world experience.
- Most courses teach you theory (what a SWOT analysis is) but leave you stranded when you have to actually apply it to a blank page.
I realized that millions of aspiring PMs are stuck in "Tutorial Hell"—watching videos but never building anything because they are afraid of doing it wrong.
The Solution: PM Portfolio Builder
I built an AI-Guided PM Workshop that acts as a "Senior PM Mentor in a Box". It doesn't just generate text for you, it forces you to think.
Unlike ChatGPT (which just gives you the answer), my app uses a specialized "Roast & Refine" engine. It critiques your answers for strategic gaps, providing you guidance to think in the right direction.
Key Features:
💡 The Idea Sandbox: Overcomes "writer's block" by generating realistic problem statements for you to solve.
🔥 The Roast Engine: A "Senior PM" AI agent that critiques your work, pointing out logical fallacies before you move to the next step.
📄 One-Click PRD Export: Automatically compiles your entire session into a professionally formatted Product Requirements Document (PDF/JSON) that you can use as a proof or work.
Demo
🚀 Try the Live App Here
📂 View the Source Code
Idea Sandbox
PM Workshop
My Experience with GitHub Copilot CLI
My experience with GitHub Copilot was absolutely amazing. I really enjoyed building the app using the CLI agent. The very first thing I did once I was clear on what needed to be built was creating an implementation plan using the Plan Mode.
I started off by listing out the idea, the problem statement, my hypothesis, and the solution, along with other relevant details required to help Copilot understand what we are building and why. The intent was to provide as much context as possible about the product. Based on these requirements, Copilot generated an IMPLEMENTATION_PLAN.md (you can view it here) which had all the details of the implementation. I then reviewed it and asked Copilot to start the implementation based on that plan.
Copilot struggled a bit with the initial setup (like installing Next.js and initial dependencies), so I manually set those up before asking Copilot to proceed. Once the foundation was laid, Copilot began the implementation and did a great job executing the plan we discussed.
Favorite Features:
Queueing Messages: One of the cool features I loved is the ability to queue messages while Copilot is working on a current task. I could add a new instruction using the enqueue feature, and once Copilot finished its current task, it automatically picked up the next one.
Session Resume: The
copilot --resume=<session_id>command is a game changer. It let me pick up work exactly where I left off, utilizing the deep context we had built up during previous sessions without having to explain everything again.
Overall, the GitHub Copilot CLI felt less like a tool and more like a pair programmer that kept me in the flow.
Top comments (0)