← All terms Module 1: The Translation Layer

Negative Prompting

cognitive-interface-architecture / negative-prompting

Definition

Explicit specification of what the model must not do. The "Guardrails" slot of the Three-Constraint Rule, treated as a first-class constraint rather than an afterthought. Negative Prompting defines the negative space that shapes the output as precisely as any positive instruction.

What this prevents

Without explicit exclusions, a model chooses the most likely completion: which may be a format you dislike, a tone you did not want, or a scope far wider than you intended.

Related terms

See all terms in Module 1: The Translation Layer →

See this term applied in production

The Agent Control Architecture Pack includes deployable system prompts, AGENTS.md templates, and fully-worked BYOP rebuilds that operationalise every precision term.

Get ACAP ($89) → Subscribe to The Constraint