Yan-Zero/dsh-progressive-tools

Plugin插件 Native原生 ⭐ 3 Apache-2.0 Prompts & Skills提示词与技能

一个自动挂载、仅用于展示的渐进式工具展示层,适用于原生、代码和双模式下的每个 DeepSeek Harness Agent 预设。

Project Overview项目介绍

This is a native DSH plugin that adds an automatically mounted, presentation-only progressive tool disclosure layer for all DeepSeek Harness agent presets across Native, Code, and combined modes. It keeps the model-visible tool prefix stable while the live ToolRuntime catalog can dynamically register, unregister, restrict, or shadow non-eager tools. You install it as a standalone bundle into your DSH profile, and it mounts automatically to all existing and future agent presets, no manual preset edits or copies required. It requires access to core DSH APIs for tools, system prompts, and LLM; codeRuntime is only needed for Code Mode agents, and it never modifies your local Harness checkout.

The plugin is designed for DSH users who need to dynamically manage large tool collections. It lets you customize limits for search results, tool description length, summary characters, query characters, tool name length, and maximum result size via your profile's configuration. You can first list all lightweight tool entries or filter by a text query with search_tools, then fetch full schemas and descriptions for only the tools you intend to use with describe_tools, then invoke the selected tools in a later model step. This workflow avoids cluttering the model's context with tools you don't use immediately, while still letting you access the full catalog when needed.

The plugin is released under the open-source Apache 2.0 license. It has several documented known limitations: it relies on pi-ai internal interfaces, so if the internal structure changes, it will fail open to a stable invoke_tool fallback. Search uses relaxed deterministic lexical ranking rather than semantic retrieval, and you should not run two plugins that both rewrite the final tool surface in the same scope. It does not modify DSH core code, so any incompatibility will not break your base Harness installation.

这是一款专为DeepSeek Harness开发的原生插件,为DSH中所有Native模式、Code模式和混合模式的Agent预设提供了渐进式工具披露层。它在允许运行时工具目录动态注册、注销、限制或隐藏非即时加载工具的同时,保持模型可见的工具前缀稳定。安装后它会自动挂载到宿主平面的所有现有和未来Agent预设,无需手动修改或复制现有预设。

适合需要动态管理工具集合的DSH用户,插件支持用户在配置中自定义限制搜索结果数量、工具描述长度、总结字符数、查询字符数、工具名字符数和结果大小等参数,还支持声明固定始终可见的即时工具列表。用户可先通过search_tools列出全部轻量工具目录或按关键词筛选工具,再调用describe_tools获取目标工具的完整元信息。

插件遵循Apache 2.0开源许可证,不会修改原DSH代码库,已知局限包括依赖pi-ai内部接口,若内部结构变更会自动降级到invoke_tool兼容方案,搜索功能是基于词汇匹配而非语义检索,同一作用域不建议同时运行多个修改工具表面的插件。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意
  • Only 3 stars - very few users, little community feedback星标只有 3,几乎没人在用,遇到问题缺少社区反馈
DSH walks through these 9 checksDSH 会逐条核对这 9 项

Compatibility兼容性

  • DSH, Node, OS and profile requirementsDSH 版本 / Node 版本 / 操作系统 / profile 是否满足要求
  • External dependencies and runtimes (Electron / Python / Docker, ...)外部依赖与运行时(Electron / Python / Docker 等)是否齐备
  • Conflicts with installed plugins: command names, skill / tool names, ports, duplicate MCP registration与已装插件是否冲突:命令名、skill / tool 重名、端口占用、重复 MCP 注册

Security安全性

  • Repo matches the facts registered here; archived or abandoned?仓库是否与页面登记一致,是否归档或长期停更
  • Safety of preinstall / install / postinstall and install.sh / setup.ps1preinstall / install / postinstall 与 install.sh、setup.ps1 是否安全
  • curl|bash, download-then-execute, obfuscation, unrelated domains → stop immediatelycurl|bash、下载即执行、混淆代码、无关域名 → 立刻停止
  • Typosquatting or unmaintained packages among the new dependencies新增依赖里有没有 typosquatting 或无人维护的包
  • Requested permissions vs. what the feature actually needs申请了哪些权限、是否超出功能所需(filesystem / network / shell / clipboard)
  • Any sudo / admin requirement, plus uninstall and rollback是否要求 sudo / 管理员权限,以及卸载与回滚方式

Anything uncertain must be marked unknown with a note on how to confirm it. This site's signal screen is a static snapshot, not a security audit.拿不准的必须标「未知」并说明要我怎么确认。本站的信号筛查是静态快照,不能替代安全审计。

Or use CLI install (for developers)或使用命令行安装(适合开发者)

CLI Install命令行安装

dsh plugin --profile web add github:Yan-Zero/dsh-progressive-tools

把 Yan-Zero/dsh-progressive-tools 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

dsh-progressive-tools

English | 中文

An automatically mounted, presentation-only progressive tool disclosure layer for every DeepSeek Harness Agent preset in Native, Code, and Both modes. It keeps the model-visible tool prefix stable while the live ToolRuntime catalog may register, unregister, restrict, or shadow non-eager tools.

Installation and composition

Follow INSTALL.md and install this standalone bundle into a dsh profile. Its bundle patch mounts the plugin once on the host plane, whose scoped ToolRuntime and SystemPrompt views carry it into every existing and future Agent preset. No preset copy, preset edit, or preset selection is required. The plugin requires tools, systemPrompt, and llm; codeRuntime is required only when an agent actually presents Code Mode. It never patches the Harness checkout.

- id: progressive-tools
  config:
    eagerTools: []
    maxSearchResults: 10
    maxDescribeTools: 5
    maxSummaryChars: 240
    maxQueryChars: 500
    maxToolNameChars: 200
    maxResultBytes: 1048576

The configuration block above is an optional override in the profile's cordis.patch.yml; the bundle already owns the progressive-tools row. eagerTools contains exact, genuinely stable tools that remain declared directly. Unknown eager names fail assembly. Every count, grapheme, and rendered UTF-8 result limit is enforced before a successful discovery result returns.

Stable mode surfaces

Effective mode Stable model-visible tools Invocation after describe
Native, native deferred API search_tools, describe_tools, eager tools ordinary tool call with the returned exact name and arguments
Native, other API search_tools, describe_tools, invoke_tool, eager tools invoke_tool({ name, arguments })
Code run_code; its SDK declares discovery and eager bindings tools[exactName](https://github.com/Yan-Zero/dsh-progressive-tools/blob/HEAD/arguments) inside a later run_code
Both run_code, discovery tools, eager tools ordinary tool call or Code binding

invoke_tool is a capability fallback, not a second Code transport. The plugin removes it at the pi-ai request boundary when the resolved model supports OpenAI Responses client tool search or Anthropic tool references. Code and Both already have run_code, so they never expose invoke_tool.

Showing the opening section of the README — the full document lives in the repository以上为 README 开头摘要,完整文档在仓库内 · View the full README on GitHub →在 GitHub 查看完整 README →

← 上一个 Prev dsh-plugin-manager 下一个 Next dsh-kb-sieve →