context-engineering-fundamentals

Context engineering fundamentals

Context engineering is the practice of managing an LLM's limited attention budget. Use this skill to keep instructions, evidence, and state available during long work.

Core concept

Context windows are constrained by attention mechanics, not only token capacity. A large context limit does not guarantee equal use of every item.

The lost-in-middle effect

The "Lost in the Middle" experiments show that retrieval quality can change with information position. The result depends on the model, task, context length, and number of documents.

Position	Common test result
Beginning	Often easier to retrieve
Middle	Can be harder to retrieve
End	Often benefits from recency

Implication: Keep critical constraints easy to find and repeat them near the decision that uses them. Do not assume position alone predicts recall.

Context degradation patterns

1. Lost-in-middle

Information in the middle of long context gets lower attention weight.

Mitigation: Structure with explicit sections. Put critical constraints at start AND end.

2. Context poisoning

Errors compound when incorrect information enters context (from tool outputs, summaries, or earlier mistakes).

Mitigation: Validate intermediate outputs. Don't blindly trust previous responses.

3. Context distraction

Irrelevant information forces attention allocation away from relevant content. Models can't "skip" irrelevant context.

Mitigation: Be selective about what goes into context. More isn't better.

4. Context confusion

Multiple task types or conflicting instructions create ambiguous responses.

Mitigation: One task per interaction when possible. Clear task boundaries.

5. Context clash

Contradictory information from multiple sources causes derailing conflicts.

Mitigation: Resolve contradictions explicitly before asking Claude to use the information.

Measure before compressing