Команды для терминала DSH
dsh plugin --profile web add dsh-context-compression-improvedЗапустите команду в терминале, где работает DeepSeek Harness. Перезапуск применит плагин автоматически.
Плагин «dsh-context-compression-improved» для DeepSeek Harness: Same-origin in mechanism with the two loudest lines in context compression. The code-skeleton gate follows the skeletonization approach of Headroom (Apache-2.0), whose published headline is 20% fewer tokens for coding agents and 60–95% fewer tokens for JSON, same answers. The estimator channel follows TokenPilot (arXiv:2606.17016), which reports up to 60% lower cost for long-session agents. Both figures are theirs, quoted as-is; this plugin ships no benchmark of its own and claims no reduction of its own. What it adds for DeepSeek Harness: choose a compression profile, set the Auto Compact trigger level and toggle code-skeleton compression from one settings section, with exact DeepSeek V4 tokenizer measurement, same-revision count verification, and fail-open behaviour that keeps the original tool results on unsupported models.
★ 2 звёзд GitHub↓ 8 184 загрузок за 30 дней18 сентября 2026 г. добавлен📦 npm verified
Что делает dsh-context-compression-improved
Плагин «dsh-context-compression-improved» для DeepSeek Harness: Same-origin in mechanism with the two loudest lines in context compression. The code-skeleton gate follows the skeletonization approach of Headroom (Apache-2.0), whose published headline is 20% fewer tokens for coding agents and 60–95% fewer tokens for JSON, same answers. The estimator channel follows TokenPilot (arXiv:2606.17016), which reports up to 60% lower cost for long-session agents. Both figures are theirs, quoted as-is; this plugin ships no benchmark of its own and claims no reduction of its own. What it adds for DeepSeek Harness: choose a compression profile, set the Auto Compact trigger level and toggle code-skeleton compression from one settings section, with exact DeepSeek V4 tokenizer measurement, same-revision count verification, and fail-open behaviour that keeps the original tool results on unsupported models.
Показать оригинальное описание (EN/ZH)
Same-origin in mechanism with the two loudest lines in context compression. The code-skeleton gate follows the skeletonization approach of Headroom (Apache-2.0), whose published headline is 20% fewer tokens for coding agents and 60–95% fewer tokens for JSON, same answers. The estimator channel follows TokenPilot (arXiv:2606.17016), which reports up to 60% lower cost for long-session agents. Both figures are theirs, quoted as-is; this plugin ships no benchmark of its own and claims no reduction of its own. What it adds for DeepSeek Harness: choose a compression profile, set the Auto Compact trigger level and toggle code-skeleton compression from one settings section, with exact DeepSeek V4 tokenizer measurement, same-revision count verification, and fail-open behaviour that keeps the original tool results on unsupported models.
Как удалить при необходимости
dsh plugin --profile web remove dsh-context-compression-improvedБезопасность и аудит
Плагины выполняются в окружении пользователя DSH. Рекомендуется проверять исходный код, зависимости и запрашиваемые разрешения в репозитории автора перед установкой.