# Vp Minimal Repro

> Reduce an observed failure to a small reproduction that fails on demand and passes once the cause is removed. Use when a failure has been seen and the next step would otherwise be to drive the whole system again to watch it, or when a reproduction has to be rebuilt standalone to demonstrate or share it. Applies at any layer, including build tools, libraries, services, runtimes, and version-specific behavior. Boundary: starts from an observed failure and ends at a re-runnable reproduction; not for filing the report or deciding where it goes.

- Skill: `vdustr/vp-minimal-repro` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add vdustr/vp-minimal-repro`
- Raw SKILL.md: https://api.skillmd.com/api/skills/vdustr/vp-minimal-repro/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Docs & Writing
- Author: VdustR (https://skillmd.com/u/vdustr)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/vdustr/vp-minimal-repro

---


# Minimal Reproduction

Turn an observed failure into something that can be run again and answers with
an exit code. Driving the whole system to watch a symptom is slow, hard to
repeat, and produces evidence nobody can re-run.

## Two levels

A **working reproduction** exists to find the problem and to test a candidate
fix. It only has to fail reliably, run fast, and be honest about what has
actually been verified. It stays where the work is and is disposable. See
`references/working-repro.md`.

A **demonstration reproduction** is an artifact built for a reader who has none
of this context. It shows the mechanism, the fix when there is one, and nothing
else, and it is designed to make that mechanism obvious rather than to recreate
the original symptom. See `references/demonstration-repro.md`.

Choose from the situation rather than from a fixed rule. While the cause or the
fix is still unknown, the answer is a working reproduction. Escalate when there
is a reason for a stranger to run it, typically an instruction to prepare a
demo or a shared artifact. If a working reproduction is about to be shared as
is, say that it can be upgraded first and what that would add.

## Requirements at both levels

Verify the reproduction yourself instead of asking anyone else to confirm it.
Run it and watch it fail; a reproduction that has only been reasoned about is a
guess. Once a candidate fix exists, also watch it pass with the fix applied and
fail again once the fix is reverted. That pair is required before claiming a
cause, not before reporting a reproduction.

Separate what was observed from what explains it. State plainly which parts are
verified by a run and which are inference, and never let an explanation inherit
the confidence of the observation it explains. When something comes from
documentation, a changelog, or an issue tracker, keep the source with it; a
demonstration reproduction carries those references into the artifact.

De-identify before a reproduction leaves the machine, and treat what survives
de-identification as a measure of whether it is minimal. See
`references/de-identification.md`.

Report where a kept reproduction lives, along with anything it installed or
started, so it is not left behind silently.

## Related skills

- [`vp-issue-investigator`](https://github.com/VdustR/skills/tree/main/skills/vp-issue-investigator)
  when the reproduction should support a problem report.
- [`vp-autodev`](https://github.com/VdustR/skills/tree/main/skills/vp-autodev)
  when the reproduced failure should proceed through a fix and PR delivery.
- [`vp-long-running-processes`](https://github.com/VdustR/skills/tree/main/skills/vp-long-running-processes)
  for processes the reproduction must keep running.
- [`vp-git`](https://github.com/VdustR/skills/tree/main/skills/vp-git) when the
  reproduction or fix should be committed.
- [`vp-session-wrapup`](https://github.com/VdustR/skills/tree/main/skills/vp-session-wrapup)
  for reporting or cleaning up reproduction artifacts.