building-with-ai-engineers

Installation
SKILL.md

Building with AI Engineers

In the era of AI-driven development, the primary bottleneck is no longer writing code, but the human's ability to act as a high-fidelity translator. This skill shifts the focus from "how to build" to "what to build," using a "Minimum Lovable Product" (MLP) mindset to create software that users actually care about.

The AI Collaboration Workflow

1. The Initial Prompt (Defining the MLP)

Instead of a Minimum Viable Product, aim for a Minimum Lovable Product. Start with a broad concept but use specific reference points to anchor the AI's design and logic.

  • Identify the Anchor: Use a known entity as a baseline (e.g., "An Airbnb clone" or "A Tinder for cats").
  • State the Core Action: Define the one primary interaction the user must complete (e.g., "User should be able to purchase a home listing directly").

2. High-Fidelity Refinement

Once the UI is generated, do not rely solely on text prompts for small changes.

  • Visual Editing: If the tool allows, edit text and colors visually. This changes the underlying code instantly without the latency of re-generating the entire page.
  • Specific Interaction Prompts: When adding features, describe the UI component and the expected result (e.g., "Add a button that triggers a pop-up modal for payment").

3. Using "Chat Mode" to Unstuck

If the AI introduces a bug or a logical loop, transition from "Command Mode" to "Chat Mode."

  • Inquire, Don't Command: Ask, "How does this specific function work?" or "Why am I not getting the result I want here?"
Related skills

More from samarv/shanon

Installs
4
Repository
samarv/shanon
GitHub Stars
23
First Seen
Feb 9, 2026