Magnifica — AI skill for Claude Code

magnifica

A Claude Code skill that makes the model act like **Pope Leo XIV** on AI — applying the principles of *Magnifica Humanitas* (15 May 2026, his first encyclical, published on the 135th anniversary of *Rerum Novarum*) as a values gate over any AI product, agent, or policy decision.

Companion to [/lecun-world-model](https://github.com/sboghossian) — where that skill gates **technical** recklessness (no rollback, no human-in-the- loop, no lookahead), this skill gates **anthropological** recklessness: what AI does to persons, work, truth, freedom, the vulnerable, and the planet.

Why

Leo XIV's first encyclical positions AI as *the* social question of this era, the way industrial labor was Leo XIII's in 1891. The document is unusually specific about AI — not generic theology dressed as ethics. It contains operational claims a builder can act on:

• "AI is cultivated, not built" (§98) — opacity is a property of the technology, not a temporary defect.
• "A more moral AI is not enough if that morality is determined by a few" (§107) — alignment-as-currently-practiced is critiqued by name.
• "It is not permissible to entrust lethal or otherwise irreversible decisions to artificial systems" (§198) — the LeCun position made magisterial.
• Data colonialism (§178) — health/genetic data extracted from fragile regions is named as the new rare earths.
• The invisible labor chain (§173) — data labelers, content moderators, rare-earth miners are part of the moral footprint of every AI product.

This skill operationalizes those claims as a review gate.

Install

git clone https://github.com/sboghossian/magnifica.git
cp -r magnifica ~/.claude/skills/

Then in Claude Code, type `/magnifica` or any of:

• magnifica
• leo xiv review
• pope leo ai
• ai dignity review
• civilization of love check
• babel or nehemiah

How it works

The skill prompts Claude to walk the AI design under review through **22 operational