Show HN: Terse, a Claude Code plugin that halves reply length by cutting filler (github.com)

A new plugin called Terse for Claude Code has been introduced, significantly enhancing reply efficiency by cutting response lengths in half and eliminating extraneous wording. Testing shows that using the Fable 5.1 model with Terse results in a 46% reduction in words and a 51% lower cost, while Opus 5.5 achieves even greater efficiency with 54% fewer words and 32% lower costs. The plugin enforces strict writing rules applicable to all types of replies, including code comments and documentation, focusing on brevity and clarity by reducing filler content like metaphors and personal pronouns. This development is notable for the AI/ML community as it addresses the growing demand for concise and cost-effective communication in programming contexts. The Terse plugin implements 22 specific writing rules aimed at optimizing response structure—demanding answers to be straightforward, concise, and devoid of unnecessary embellishments. With significant performance improvements shown in measured trials, Terse not only decreases the number of output tokens but also minimizes style violations, suggesting a substantial step forward in natural language processing applications for coding and technical tasks.