DSH Recap:在 Harness 会话里按需回顾任务进度

dsh-recap 是 lolipop-1x1 为 DeepSeek Harness 开发的记忆类插件,通过 /recap 与可选的三分钟空闲自动摘要,在当前会话内简要说明任务、进度与下一步。支持模型与语言设置、后台生成与缓存复用。安装时在 Plugins 粘贴 GitHub 仓库 URL 并从源码构建;兼容 Harness 0.2.0-rc.2,需 Node.js ≥22.19。本文基于官方 README 与 package.json 整理安装、命令与默认预算,并提醒安装前审阅脚本与进程权限。

Read More
DeepSeek Harness Plugin: dsh-codex-compaction

dsh-codex-compaction is a plugin for DeepSeek Harness designed to address the long-context bloat issue in Codex sessions. While preserving the official Basic Compaction mechanism, it provides an optional account-level native summarization interface, image replay fixes, and legacy compatibility support. Users can independently enable or disable native compaction in a session via a command. The plugin supports custom models and does not modify the core code. Installation requires a specific version of the DSH CLI, and a restart is needed for it to take effect. When in use, note the native retry limit and potential information loss in semantic recall.

Read More
dsh-session-intelligence: Right Sidebar Session Summary and Git Status

The `dsh-session-intelligence` plugin aims to address the issue of forgotten goals in long DSH sessions. It displays the session’s original intent, a summary of recent user requests, runtime status, and read-only Git information in the right sidebar. This plugin depends on `dsh-better-sidebar` and requires a local web environment to run. The initial summary generation takes about 20 seconds, and summaries are refreshed approximately every 30 seconds. During installation, make sure to remove any outdated conflicting plugins and ensure that the Node.js version and dependency versions meet the requirements. Do not expose it to the public internet through a reverse proxy during use, and note that the summarization feature may incur additional costs.

Read More
DeepSeek Harness (DSH) Extension: dsh-session-summarizer

The native DSH ecosystem plugin `dsh-session-summarizer` is designed to address context loss in long conversations. It does not rely on external services and uses the official DSH runtime to read sessions and generate Chinese summaries, helping models quickly recover context. The plugin can be invoked via the `session_summarize` tool, allowing users to specify a session ID and the number of turns to summarize. Its transcription strategy retains the beginning and end while truncating the middle, with a default budget of 12,000 characters. After installation, models can directly call this tool to manage session summaries.

Read More
dsh-memory: Long-term Memory Plugin for DeepSeek Harness

dsh-memory is an MIT-licensed long-term memory plugin for DeepSeek Harness (DSH), designed to address the lack of a stable medium for user preferences, verification steps, and failure lessons in multi-session workflows. Distributed via GitHub, it utilizes structured extraction to convert reusable information from completed sessions into summaries and memory files, and automatically injects them into subsequent sessions. Core features include: debounced extraction after session completion (default 3 minutes), global cooldown integration (default 6 hours) into MEMORY.md and memory_summary.md, persistent summary injection, and the provision of four query and add tools: memory_list, memory_read, memory_search, and memory_add. Among them, memory_add is only used when...

Read More