Skip to main content

In this article

RPI Walkthrough

Use /rpi-walkthrough when you want to understand code or an artifact before deciding what to change. The walkthrough traces the target first, divides it into coherent segments, and explains one segment at a time with links to the relevant evidence.

The skill is read-only by default. It records a decisions-and-changes ledger only when you make a material decision or request a change during the conversation.

Start a Walkthrough​

Invoke the skill with a target and optional detail level:

/rpi-walkthrough target=README.md detail=deep

You can also use an open or attached file as the target:

/rpi-walkthrough detail=normal chat

The available inputs are:

InputPurpose
target=...Select a file, feature, UI or UX area, library, or .copilot-tracking artifact
detail=briefFocus on purpose, structure, and the main flow
detail=normalBalance the main flow with important implementation details; this is the default
detail=deepTrace detailed behavior, dependencies, and design rationale
chatUse the current conversation to refine the target and starting point

When no target can be inferred, the skill asks what you want to explore. When several unrelated targets match, it asks you to choose one before continuing.

Follow the Segment Loop​

Before the explanation begins, the walkthrough reviews the target and plans a sequence that follows its real structure. A code walkthrough might move from an entry point through control flow and key dependencies. An artifact walkthrough usually follows its sections in reading order.

Each segment includes:

  • A heading that shows your current position
  • An explanation of what the segment does, how it connects, and why it exists
  • Inline links and a reference table pointing to the relevant files and lines
  • One or two questions that let you continue or request more depth

Before the first segment, the walkthrough includes an overview diagram when the target has meaningful structure or flow. A later segment includes a focused diagram only when it clarifies relationships not already clear from the overview and prose.

The skill pauses at each segment boundary. You control the pace and can change the detail level during the walkthrough.

Capture Decisions and Requested Changes​

An explanation alone does not create an artifact. When you make a material decision or request a change, the skill creates a dated ledger at:

.copilot-tracking/walkthroughs/{{YYYY-MM-DD}}/{{task_slug}}-decisions.md

The ledger keeps the request connected to the evidence that prompted it. Each entry is reconciled as applied now, handed off to RPI work, deferred, or declined. The walkthrough does not modify source files unless you explicitly ask it to apply the requested change now and the change is safely scoped.

For example, after learning how an authentication path handles expired tokens, you might request consistent audit logging. The walkthrough records that request and its disposition without interrupting the remaining explanation.

Continue into RPI Work​

A walkthrough may use /rpi-research to gather external evidence needed for an accurate explanation. In a standalone walkthrough, it does not automatically continue into downstream RPI work. At closeout, it recommends the smallest applicable RPI command when a ledger entry was handed off or still needs further work:

NeedRecommended command
A decision depends on missing evidence/rpi-research
A requested change needs a durable implementation strategy/rpi-plan
A scoped, approved change is ready to apply/rpi-implement
Completed work needs acceptance review/rpi-review
Several lifecycle stages need coordination/rpi-quick

When an active RPI parent workflow owns continuation, the walkthrough returns the ledger and evidence to that parent instead of selecting the next command. When no decision or requested change needs follow-up, it closes without an RPI handoff.

Next Steps​

🤖 Crafted with precision by ✨Copilot following brilliant human instruction, then carefully refined by our team of discerning human reviewers.