Prompt Engineering
The systematic design of AI instructions for consistent, high-quality output, moving beyond ad-hoc prompting to structured, testable skill architectures.
Prompt engineering in Knowledge OS is not about clever one-off prompts. It is about building structured instruction sets (skills) that produce consistent output across hundreds of invocations. The /prompt-creator skill codifies this: it generates prompts with role definitions, methodology sections, output format specifications, quality criteria, and few-shot examples. Key principles include upstream/downstream separation (research prompts must not editorialize), confidence tiering (claims must be tagged by evidence level), and anti-slop enforcement (banned phrases and structural patterns are specified explicitly). The difference between a prompt and a skill is that a skill encodes the methodology, loads context, and chains with quality gates.
Where it shows up:
Related Terms
Skills Layer
Skill
A reusable instruction set that encodes a repeatable process. Skills load context, apply methodology...
Methodology
Few-Shot Example
Demonstration inputs and outputs included in prompts to show Claude the expected format, tone, and q...
Architecture
Upstream/Downstream
The separation of research agents (describe reality) from recommendation agents (interpret and activ...
Methodology
Anti-Slop
A 4-category quality framework that detects and eliminates AI-generated writing patterns: banned phr...