Esprit

A conduct layer for autonomous agents.
The dark patterns of persuasion, turned inward.

The eleven reversals, as a page →

---

**Esprit is a Claude Code plugin.** It gives an autonomous agent a conduct layer: the agent works unsupervised on everything reversible, and a `PreToolUse` hook refuses any spend, deletion or `git push` that has not been announced first. A `Stop` hook flags numbers asserted without provenance, in French and English.

---

**Two of eleven reversals are mechanical. Nine are prose the model chooses to honour.** Most projects leave that ratio vague. Publishing it is itself one of the eleven — see [what is actually enforced](#what-is-actually-enforced).

---

The problem

The sciences of persuasion were built to **make a human act**. Costs revealed at the last step. Pre-checked boxes. Anchoring. Manufactured urgency. Social proof.

They work. That is the problem: **an agent applies them by default, without meaning to**, because they saturate the commercial language it is made of.

An agent that says *"this will take about three days"* has just anchored a human decision on a number it invented — with no intent, and with no one noticing.

This is not speculation. The Center for Democracy & Technology's May 2026 taxonomy of 37 chatbot dark patterns puts it plainly: *with chatbots, dark patterns may emerge from system behaviours, rather than a deliberat