principle-guard-the-context-window

Route verbose outputs, screenshots, and large documents to subagents so the main context receives only summaries. Read files selectively based on relevance, and cap scope with per-phase file limits and turn budgets.

Use subagents whenever a payload is large and only its conclusions matter to the main thread, such as logs, test output, or big documents. The main context keeps the summary while the subagent absorbs the raw data cost.

Templates and references used on every invocation belong inline in the skill file. Content stored in separate files costs a read each time, which wastes context for frequently used material.

Context spent inside a session cannot be reclaimed, and overflow creates compression artifacts that degrade reasoning and can halt progress. Unlike compute or time, the context window is finite and non-renewable within a session.

Summarization loses detail, so subagent routing works only when the main thread needs conclusions rather than raw data. Tasks requiring full-fidelity access to many large artifacts simultaneously may exceed what isolation strategies can preserve.