Prompt engineering

Deliberately encoding rules, voice, and constraints into the instruction text fed to a model to steer its behavior without retraining.

grounded in: principles.md is loaded verbatim into the self-improve/idea-engine prompts, and recent commits (6af639d, ecc99cf) bake explicit 'Growth Editor' editorial rules into the LinkedIn generator prompt

Connected concepts

Context engineering, Agentic-SDLC doctrine, Idea engine, Self-improvement daemon, AI-slop / de-slopping