Automated summary

Runs a supervised dry-run of a prose skill in a disposable sandbox to validate its output without affecting real areas.

Strengths

Validates generative skills safely without production impact
Uses structural invariants and golden snapshots for robust testing
Catches regressions automatically when refactoring or modifying skills

Limitations

Not suitable for conversational skills with non-deterministic output
Requires a pre-built test kit (golden_inputs, verify.sh) to be effective
Can be overhead-heavy for very complex skills

When to use it

Before modifying or refactoring any skill that produces verifiable artifacts (files, scaffolds, configs).

When not to use it

For purely advisory or conversational skills where responses are unpredictable or cannot be snapshot-tested.

Security analysis

Safe

Quality score82/100

The skill describes a sandboxed dry-run procedure that explicitly avoids side effects, real-world changes, and destructive actions. No risky commands or data exfiltration are instructed.

No concerns found

How this analysis is made

Examples

Validate /project skill

/dry-run-skill project

Dry-run after refactoring

Run a dry-run on the refactored version of /discovery skill against its golden inputs.

Check new generator skill

Perform a dry-run of the /scaffold skill using its test kit in /tmp/dryrun-scaffold-1.

/dry-run-skill — Valida uno skill in prosa senza eseguirlo in produzione

Name: Skill dry-run validation
Author: Skills Guide Bot

Quando ricevi questo comando (con il nome di uno skill, es. /dry-run-skill project), esegui un dry-run sorvegliato di quello skill: lo lanci in una sandbox usa-e-getta con input fissi, ne verifichi gli INVARIANTI dell'output, e riferisci PASS/FAIL — senza toccare aree reali.

Perché esiste: gli skill in prosa (/project, /discovery, generatori di scaffold/file) li esegue l'LLM — non hanno un test runtime nativo. Rifattorizzarli o modificarli "al buio" rompe in silenzio. Questa è la rete. Dettaglio e fondamento: SSOT DRYRUN_PROTOCOL e KNOW_HOW_TRANSFER_PROTOCOL (paradigma).

Quando usarlo

PRIMA di modificare/rifattorizzare uno skill che genera artefatti verificabili (file, scaffold, descrittori, config).
Per validare uno split/refactor (es. /project monolitico → micro-skill): si confronta il nuovo output col golden.
Come test di accettazione di uno skill nuovo generatore.
NON serve per skill conversazionali/consulenziali (niente output deterministico da snapshottare).

Il "kit" di dry-run di uno skill (convenzione)

Ogni skill validabile ha un kit con tre pezzi (è la convenzione da seguire/creare se manca):

golden_inputs.json — risposte canoniche FISSE, scelte deterministiche e senza side-effect (per /project: paradigma-solo, niente librerie, RAG offline).
verify.sh <dir> (o verify_scaffold.sh) — verifier degli INVARIANTI strutturali (NON byte-diff): placeholder risolti, campi-chiave corretti, schema atteso, ecc. Deve essere una rete vera (bocciare un albero rotto).
golden/<...> — snapshot dell'output dello skill ATTUALE con quegli input (riferimento d'oro).

Kit di riferimento esistente per /project: os3-matrix/docs/tests/m-os3-065/.

Procedura (eseguila così)

Localizza il kit dello skill richiesto (golden_inputs.json + verify*.sh + golden/). Se manca, dillo e proponi di crearlo (non inventare input).
Sandbox: crea una dir usa-e-getta /tmp/dryrun-<skill>-<n>. Lavora SOLO lì.
Esecuzione sorvegliata: spawna un agente (o esegui tu) con istruzione di ESEGUIRE lo skill target (leggi il suo .md) contro la sandbox usando SOLO golden_inputs, in modalità non-interattiva (non chiedere nulla) e senza side-effect (NON clonare, NON installare, NON toccare aree reali).
Verifica: lancia verify*.sh <albero-prodotto> → exit 0 atteso. Se è un refactor, confronta anche i file chiave col golden/ (match semantico, non byte).
Riferisci: PASS (zero violazioni) o FAIL con l'elenco esatto degli scostamenti.
Teardown: rimuovi la sandbox.

Esito

PASS = lo skill (o la sua versione rifattorizzata) produce un output conforme agli invarianti e allineato al golden.
FAIL = scostamenti da spiegare e correggere PRIMA di attivare/spedire la modifica. Mai attivare al buio.

Vedi anche: agente skill-dryrun-guardian (si auto-attiva quando stai per toccare uno skill in prosa e ti ricorda di usare questa rete), KNOW_HOW_TRANSFER_PROTOCOL (perché questa capacità è nel prodotto, non in memoria privata).

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/skill-dry-run-validation-mrwake1t/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/skill-dry-run-validation-mrwake1t' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/skill-dry-run-validation-mrwake1t/download' --output '.skill-downloads/skill-dry-run-validation-mrwake1t/SKILL.md.part' && mv '.skill-downloads/skill-dry-run-validation-mrwake1t/SKILL.md.part' '.skill-downloads/skill-dry-run-validation-mrwake1t/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

This file does not provide a usable name/description header for this native procedure. It remains available to download and inspect; use the manual method below.

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/skill-dry-run-validation-mrwake1t/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.