Skip to main content

In this article

rpi-challenger

rpi-challenger
FieldValue
Kindskill
Source.github/skills/rpi/rpi-challenger
InvocationInvoked directly as /rpi-challenger, or loaded on demand by referencing agents
InteractiveNo

What it does

Challenge a confirmed task, decision, plan, or artifact through adaptive skeptical questions. Use when you need to expose assumptions before acting.

When to use it

Use rpi-challenger before committing to a task, decision, plan, or artifact when you want its assumptions exposed rather than confirmed. The skill first proposes a factual scope and waits for your confirmation, then asks one open-ended, non-leading question per turn, following your answers to the most material remaining uncertainty.

It records the exchange, coverage, and unresolved items in .copilot-tracking/challenges/{{YYYY-MM-DD}}/{{task_slug}}-challenge.md and leaves every other artifact untouched.

During the active exchange it does not solve, review, validate, or recommend. At closeout it summarizes coverage and unresolved items and may name an /rpi-* command as an advisory next step, without invoking it.

Reach for a different asset when:

Example usage

/rpi-challenger subject="Move all pipeline writers to the async storage client" artifacts=.copilot-tracking/plans/2026-09-04/blob-storage-plan.md focus="rollout risk"

The skill states the candidate scope and asks you to confirm it, then begins:

Scope: the P02 writer migration in the blob-storage plan, focused on rollout risk. Confirm to begin.
What evidence tells you the existing synchronous writers can be switched without changing
their callers' error handling?

After you end the session, the closeout separates session status from what remains open:

* Session: Complete; 4 angles explored, 2 unresolved
* Unresolved: caller error-handling contract (needs code evidence); rollback path if the async client fails in production (needs a decision)
* Advisory next option: `/rpi-research` for the caller contract before P02 begins