dsh-skill-router
On-demand skill tools for DeepSeek Harness: registers skill_search / skill_load / skill_ref so the model searches a staged skill library first and loads a skill body only when it needs it, keeping the library out of the resident session catalog and out of every turn's context; adds a Skills tab to the conversation view ring that reads back the whole session and lists which skills were actually loaded, and in which turn. Zero dependencies, and the records render client-side only, never entering the model's context.
DeepSeek Harness 的按需技能工具：注册 skill_search / skill_load / skill_ref 三个工具，让模型先从分层的技能库里检索、再按需加载正文，库里的技能因而不必进入常驻目录、不占每轮上下文；并在对话视图栏加一个「技能」标签页，回读整个会话、列出本次真正加载过哪些技能及其轮次。零依赖，所有技能调用记录只在客户端渲染，不进入模型上下文。
How to install
dsh plugin add dsh-skill-router About
dsh-skill-router English | 中文 给 DeepSeek Harness 的 **Host 插件**。新增 skill_search / skill_load / skill_ref 三个工具，让 agent 在需要时自己去技能库里检索并加载技能。 **你只有十几个技能？这个插件不适合你。** 一次工作通常只用到几个技能，所以**技能少的时候，让它们常驻目录反而更省**——目录本来就是 DSH 的原生机制。 这个插件解决的是另一个问题：**技能多到装不进目录**。 你需要它吗 判断标准只有两条：**库有多大**，以及**每个任务用几个**。 | 你的情况 | 怎么办 | |---|---| | 技能 < **~30** 个 | **不需要本插件。** 全部常驻，目录成本可控，原生 skill 工具直接能用 | | 技能几十到上百，且每个任务只用几个 | **本插件的典型场景。** 常驻留高频的十来个，其余进库按需检索 | | 有多个上游仓库、上千个技能 | **最需要。** 全量常驻不可行（1000 条 ≈ 每轮 15 万字符），但你又不想丢掉其中任何一个 | | 只想"技能自动触发"、库其实很小 | **不需要。** 那是 pre-step 路由插件的领域，不是这个 | 一句话：**它是为"库大、但每次只用几个"设计的**。如果你把常用技能都常驻了，它的收…
Recommendation signals
Meta
- License
- MIT
- Language
- JavaScript
- GitHub stars
- 0
- mo. downloads
- 100
- Last push
- 2026-09-24
- Created
- 2026-09-23
Links
Basic safety check
- Findings
- None
- Sources
- curated:awesome-dsh-plugin.com, curated:awesome-dsh-plugin/awesome-dsh-plugin
- Topics
- deepseek-harness, dsh-plugin
Related plugins
superpowers
obra/superpowers
An agentic skills framework & software development methodology that works.
dev-browser
SawyerHood/dev-browser
A Claude Skill to give your agent the ability to use a web browser
ruflo
ruvnet/claude-flow
🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated