rpi-walkthrough

Explains code, UI, and artifacts through guided conversational segments with evidence links.

1.4k|284|Updated Nov 2, 2025
One-click install
npx skills add https://github.com/microsoft/hve-core --skill rpi-walkthrough
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: rpi-walkthrough
Source: https://github.com/microsoft/hve-core/tree/main/.github/skills/rpi/rpi-walkthrough
Command: npx skills add https://github.com/microsoft/hve-core --skill rpi-walkthrough

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Understanding unfamiliar code, UI flows, or project artifacts often requires hours of manual tracing. This Skill walks you through any target one segment at a time, grounding every explanation in deep subagent review and navigable file-and-line links so you learn how and why something works without reading every line yourself.

Core Features & Use Cases

  • Segmented Explanations: Breaks a target into coherent segments with headers, Mermaid diagrams, inline markdown links, and reference tables, pausing for your direction between segments.
  • Deep Subagent Review: Dispatches exploration subagents and optional research passes before explaining, so explanations are grounded in actual code behavior rather than guesses.
  • Decisions-and-Changes Ledger: Lazily captures material decisions or requested changes in a dated .copilot-tracking ledger and reconciles them as applied, handed off, deferred, or declined.
  • Use Case: You inherit a feature you did not write. Ask for a walkthrough of its entry point and data flow, and receive a paced, linked explanation with an overview diagram, then record a requested refactor in the ledger for later RPI follow-up.

Quick Start

Ask the assistant to walk you through a specific file, feature, or .copilot-tracking artifact at normal detail, one segment at a time.

Frequently Asked Questions about rpi-walkthrough

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I get an AI to explain unfamiliar code step by step?

Invoke the walkthrough with a target file, feature, or artifact and a detail level of brief, normal, or deep. It reviews the code with exploration subagents, then explains one segment at a time with diagrams, inline links, and a reference table, pausing for your input between segments.

What targets can the rpi-walkthrough skill explain?

It handles source code, UI or UX wiring, libraries and features, prompt-engineering artifacts such as prompts, instructions, agents, and skills, and .copilot-tracking artifacts like research, plan, changes, review, or log documents.

Does the walkthrough modify my source code?

No, it is read-only by default and only explains and captures decisions. Source changes happen only when you explicitly choose immediate reconciliation of a ledger entry and the change is safely scoped.

When does the walkthrough create a decisions-and-changes ledger?

The ledger is created lazily only when you make a material decision or request a change during the session. It is stored under .copilot-tracking/walkthroughs with the date and task slug, and each entry is later reconciled as applied, handed off, deferred, or declined.

Can I change the explanation depth during a walkthrough?

Yes, the detail level can be switched between brief, normal, and deep at any segment boundary. Asking for more depth triggers a fresh subagent review before the segment is re-explained.