yin52133/dsh-luban--packages-dsh-luban-context ↗★ 0

@yin52133/dsh-luban-context

Auditable context compaction and virtual-file replay for DSH

AI Analysis

核心用途是在长对话中自动压缩上下文并生成虚拟文件摘要，以降低 Token 消耗。适合进行超长对话或大型开发任务的用户。需要配合 telemetry 提供者使用。

Development

Package: @yin52133/dsh-luban-context
Version: 0.1.1
License: MIT
Last updated: Sep 1, 2026

Install

npx -p @deepseek-ai/dsh dsh plugin --profile web add github:yin52133/dsh-luban#fd42aad0a6c5628a3bee24796f0caba131fb0252&path:packages/dsh-luban-context

README

Read the full README ↗

Configuration

- insert:
    - id: luban-context
      name: @yin52133/dsh-luban-context
      config:
        trigger: { ratio: 0.80, minGapRounds: 4 }
        strategy: summarize+virtualfile
        keepRecentTokens: 24000
        archiveDir: .luban/context-archive
        nightProfile:
          trigger: { ratio: 0.70 }
          keepRecentTokens: 16000

archiveDir must be relative to the session workspace. An unknown telemetry ratio does not trigger compaction.