Type: url
Original: https://www.philschmid.de/context-engineering
Status: partial
Summary
Philipp Schmid’s June 2025 blog post argues that the decisive skill in building AI agents is shifting from prompt engineering to context engineering, which it defines as designing dynamic systems that provide an LLM with the right information and tools, in the right format, at the right time. It lists the components of context (instructions, user prompt, short-term history, long-term memory, retrieved information, tools, structured output), illustrates the difference with a meeting-scheduling example contrasting a “cheap demo” agent and a “magical” one, and claims that most agent failures are now context failures rather than model failures.