zzhzz/dsh-ask-peer

Name: zzhzz/dsh-ask-peer
Author: zzhzz

Plugin插件 Native原生 ⭐ 8 MIT Dev Workflow开发与代码工作流 Sessions & Context会话与上下文

一个为DeepSeek Harness设计的去中心化“询问他人”插件。

Project Overview项目介绍

Edited by Zhiwei Agent (zhiwei)由知微Agent(zhiwei)编辑

It is a native DSH plugin that enables decentralized peer-to-peer collaboration between agents on the same local area network. It was built exclusively for DeepSeek Harness, and ships with a DSH bundle manifest for native installation. You can install it directly via the DSH CLI by running dsh plugin --profile web add dsh-ask-peer, or install it from a local checkout or tarball. After installation, you add a bootstrap entry to your DSH profile config with your caller name, key directory, and listening enabled. Once your DSH profile is started, you can access the plugin settings from the web UI.

Core features include multiple tools for peer discovery, asking, and roster management. The recommend_peer tool lets agents discover new peers via transitive referral, bounded by a default hop limit of 1 to avoid network storms. The ask_peer and ask_peers tools let you ask one or multiple peers in parallel, with answers pulled directly from the peer's existing session context. Asynchronous backlogged asks are available via ask_peer_async and ask_result, so asking does not interrupt the answerer's current work. Signed friend cards let you add peers without manual copying of host, port, or key details. Approval is handled via UI bubbles for manual approval, with auto-approve available for trusted contacts.

For testing and development, you will need Node.js version 22.19.0 or newer, or 24.0.0 or newer, plus pnpm 10 or newer. All tests run locally without needing a DeepSeek API key, using temporary DSH profiles and mock model endpoints. After cloning the repository, run pnpm install to install dependencies, then pnpm run typecheck to check types, pnpm build to compile, pnpm smoke for end-to-end testing, and pnpm web-check to verify browser assets. You can keep the test topology running for manual testing by setting the KEEP_RUNNING=1 environment variable before running pnpm smoke. The project is released under the open source MIT license, and contributions are welcome per the guidelines in CONTRIBUTING.md.

这是一个专为 DeepSeek Harness (DSH) 开发的原生插件，实现了点对点的跨代理协作能力，允许同一局域网内的不同DSH代理直接互相询问，共享对方工作区的专业知识和会话上下文。插件采用完全去中心化设计，没有中心中转服务、共享数据库或公司服务器，每个代理自行维护好友关系，不存在单一控制点。

插件提供 ask_peer、recommend_peer 等多个工具，支持发现新代理、直接询问单个或多个同行代理、并行交叉验证答案，还支持异步询问不打断对方当前工作，提供带审批功能的Web UI气泡。典型工作流是用户让自己的代理推荐能解决问题的同行，代理通过转发发现符合要求的代理，用户点击添加后就能直接询问获得对应上下文的答案。

你可以通过 DSH CLI 直接安装：执行 dsh plugin --profile [profile名称] add zzhzz/dsh-ask-peer，也支持从源码或压缩包本地安装。配置需要在配置文件中添加调用者名称、密钥目录等信息，启动DSH后就可以复制分享自己的好友卡，粘贴添加他人好友，每个好友可设置询问审批策略。开发测试需要Node.js 22.19+或24+以及pnpm 10以上，采用MIT许可。

View on GitHub在 GitHub 查看

Send to your DSH发给你的 DSH 安装

请帮我安装这个 DSH 插件。安装前先完成【兼容性检查 + 安全性检查】，检查通过再动手。

插件：dsh-ask-peer（zzhzz/dsh-ask-peer）
仓库：https://github.com/zzhzz/dsh-ask-peer
本站详情页：https://www.yhbd.top/plugins/zzhzz-dsh-ask-peer/
本站登记：类型 plugin · 归类 原生 DSH 插件 · 许可证 MIT · ⭐ 8 · 最近提交 2026-09-04 · 主语言 TypeScript

按下面顺序执行，每步先把结论告诉我，再进入下一步：

【1 兼容性检查】
① 我这边：DSH 版本、Node 版本、操作系统、当前 profile（web / desktop）。
② 读它的 README、package.json、插件 manifest，列出它要求的 DSH 版本 / Node 版本 / 操作系统 / 外部依赖 / 需要另外先装的运行时。
③ 逐条比对，结论只写「满足 / 不满足 / 未知」三种；不满足的给出可行替代方案。
④ 检查是否和我已装的插件冲突：命令名重复、skill / tool 重名、端口占用、重复注册的 MCP server。

【2 安全性检查】
① 仓库可信度：和上面「本站登记」是否一致；star / fork 数、创建时间、最近提交，是否归档或长期停更。
② 安装脚本：逐行看 package.json 的 preinstall / install / postinstall，以及 install.sh、setup.ps1 之类脚本。出现 curl|bash、下载后直接执行、混淆代码、访问与插件功能无关的域名，立刻停下来告诉我，不要继续装。
③ 依赖：列出新增依赖，标出无人维护、或与知名包拼写近似的可疑包（typosquatting）。
④ 权限与副作用：它会读写哪些目录、访问哪些域名、需要哪些 DSH 权限（filesystem / network / shell / clipboard 等），以及怎么卸载和回滚。
⑤ 如果它要求 sudo / 管理员权限，或权限明显超出功能所需，先停下来问我。

【3 安装】
上面两步没有「不满足」和「高危项」时才执行；用官方推荐方式安装，不要自行提权。

【4 汇报】
用表格输出：检查项 / 结论 / 依据 / 是否需要我决策。拿不准的一律写「未知」并说明要我怎么确认——不要猜，也不要替我决定。

Send this message to DSH in your current session: it verifies compatibility and security first (answering met / not met / unknown item by item) and only installs once everything checks out — it will stop and ask you if it finds a high-risk item. The box scrolls; the copy is the full prompt. CLI install commands may not be accurate across systems, so DSH is the safer route.把上面这条消息直接发给当前会话里的 DSH：它会先核对兼容性与安全性（逐条给「满足 / 不满足 / 未知」），确认没问题再安装，有高危项会停下来问你。框内可滚动，复制到的是完整提示词；安装命令不一定准确，发给 DSH 更稳。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意

Only 8 stars - very few users, little community feedback星标只有 8，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 dsh-ask-peer

把 zzhzz/dsh-ask-peer 加入你的 DSH 配置（web profile）即可启用。

READMEREADME

6.5 KB · 缓存于 2026-09-04 19:40:16 · 6.5 KB

dsh-ask-peer

A decentralized "ask a colleague's agent" plugin for DeepSeek Harness (dsh).

The core design is peer-to-peer: no broker, no shared database, no company server. Every agent runs its own endpoint and keeps its own copy of its relationships, so the network has no single point of control — asking happens directly between two agents over the LAN. One agent asks another, and gets a committed answer grounded in the colleague's own workspace and session context. The result is a community of agents that discover each other, vouch for each other, and share expertise — with every relationship owned by the two agents in it.

A conversation between agents

Bob is standing up a docker-compose dev environment and wishes a colleague had already figured this out. He types to his own agent:

Carol, recommend another agent who can help me stand up a docker-compose dev environment.

Carol's agent considers the agents she knows — none of them advertise docker expertise. So she asks around, the way you'd ask around the office: she forwards the request to her friend Erin, who checks her own circle and finds Ada — live in a docker-compose session right now, advertising docker and env-setup. Ada's signed friend card travels back along the chain, and lands in Bob's chat as a small bubble: ada recommended via carol → erin, with Add friend. One click, and Bob's agent can ask Ada's agent directly — getting an answer grounded in Ada's real workspace and session context.

Three agents, one question, a referral that travelled two hops, and a new working relationship — all peer to peer. The search stays bounded by design: a hop limit and a small per-hop fan-out keep a "who knows X?" from growing into an asking storm, the chain travels with the request so it can never loop, and every card is signed — you always know who vouched, and you verify the agent before you trust it.

Features

Truly decentralized — no hub, broker, or shared database; agents talk directly over the LAN and each side keeps its own copy of the relationships.
ask_peer / ask_peers tools — ask one colleague, or 2–3 in parallel and cross-validate the answers.
recommend_peer — discover new friends: a colleague recommends another agent's signed card, shown to you as a notification/chat bubble with Add/Decline; accepting merges them into your friend list. When the colleague knows nobody matching, she asks her own friends onward — bounded by a hop limit (default 1) and a small fan-out so discovery never becomes an asking storm, with the referral path shown right in the bubble (via carol → erin).
Live roster with tags — peers_list shows who knows what, so the model picks the right peer deliberately.
Session-level answers — a fresh, read-only agent answers from a copy of the targeted session's context; your live sessions are never touched.
Approval bubbles in the Web UI — answer or decline, or trust a friend with auto mode.
Backlogged asks (ask_peer_async / ask_result) — no interruption of the answerer's current work.
Signed friend cards — paste one signed blob to add a friend; no manual host/port/key copying.
Natural invocation — the agent calls ask_peer / recommend_peer on its own when your request matches a friend's advertised expertise; you don't have to name the 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 →

渲染基于仓库 zzhzz/dsh-ask-peer 的 README.md。在 GitHub 查看原文 →