Context Efficiency¶
Don't re-read the entire repository¶
There is no need to load every document in this repository for every task. Use context-loading-order.md to determine the specific sequence relevant to the task at hand, and stop there unless something in that sequence points to a further document you actually need.
Prefer targeted reads¶
Read the specific section or document that answers the question in front of you rather than loading large documents wholesale "just in case." If a document is long, look for its index/table of contents or headings first, then read the relevant part.
Summarize rather than duplicate¶
When referencing another document's content, link to it and summarize the relevant point rather than copying large blocks of its text into a new document. Duplicated content drifts out of sync over time and violates the "single source of truth" principle that runs throughout this repository's governance (see docs/governance/authority-and-inheritance.md).
Why this matters¶
Efficient context use isn't just about speed — over-loading context increases the chance of conflating draft/example content with authoritative content, and duplicated text is one of the most common sources of silent contradiction between documents (see prohibited-behaviors.md, item 21, on redundant instruction files).