ruleset-optimization

Ruleset optimization minimizes context usage by extracting procedural content into separate skills and applying progressive disclosure. This keeps your main ruleset lean, reduces token consumption per session, and lets Claude focus on relevant instructions for each task.

Extract procedural how-to sections, workflow guides, and domain-specific instructions that aren't needed for every interaction. Use the extraction decision tree to identify high-token, low-frequency content; auto-activating skills load only when relevant, cutting unnecessary context overhead.

Token savings calculation quantifies baseline and session-specific reductions by analyzing what content gets extracted into skills. The formula measures context removed multiplied by activation frequency, showing tangible performance gains from a leaner ruleset structure.

Yes. Ruleset optimization covers deduplication rules that identify and consolidate repeated instructions across personal CLAUDE.md, project settings.json, and skill files, eliminating redundant context and reducing overall token footprint.

Structure extracted skills with activation patterns tied to task context or keywords so they load automatically when needed. Ruleset optimization provides skill-structure guidelines and anti-patterns to follow, ensuring extracted content integrates seamlessly without manual intervention.

Ongoing maintenance involves monitoring which extracted skills activate most, periodically reviewing CLAUDE.md for scope creep, and reapplying deduplication rules. This keeps your ruleset efficient and prevents context bloat from accumulating across updates.