model-matchmaker

Monitor conversation context and prevent MAX mode by warning at token thresholds and generating handoff summaries. Use when context approaches 100K/150K/180K tokens or when working with high-cost files.

Analyze your Model Matchmaker override patterns and tune the local classifier to match your preferences. Use after collecting 50+ recommendations.

Build a sanitized analytics report from your Model Matchmaker usage for community contribution. Optional; use optimize-classifier for private tuning instead.