Context engineering patterns
Source inspected
Give the model enough context, not the whole archive.
I select the context for each task, check whether the sources are sufficient and require a structured output the next role can inspect.
01 THE PROBLEM
What needed to change
Operational archives contain inconsistent and irrelevant material. Passing everything into a prompt obscures requirements and supporting facts.
02 THE IMPLEMENTATION
What I built and directed
I developed paths that assemble requirements, criteria, risks and selected prior material. Project memory and bounded chat history preserve continuity. Source-linked snippets support lexical retrieval. Sufficiency checks and output validation make missing context and malformed results explicit.
03 THE CHECK
What the evidence establishes
Context and generation code inspected across separate applications. The implementations preserve task requirements, source sufficiency and structured output contracts.
Scope of this example
Lexical retrieval, context sufficiency and structured-output validation are implemented in separate applications.
04 THE LESSON
The judgement behind the code
Context engineering is a selection and responsibility problem. Keep the source of a fact visible after generation.
Public references
I can walk through a sanitised example without sharing private source archives or client material.