dsh-feishu

中文 · English

中文

DeepSeek Harness 的飞书前端。 没有网页，没有 TUI，没有审批弹窗。模型、斜杠命令、进度、Goal、会话，全部发生在飞书里。

dsh 官方默认面是 Web。这套插件把它换成一条长连接：进程常驻，飞书就是整个产品。

为什么是飞书，不是再做一个网页

写代码的 Agent 已经很强。缺的是一个你每天都在用的入口。

飞书不是旁路通知，是唯一交互面：

斜杠命令在飞书里敲，效果跟终端一样
模型热切换当场生效，下一轮就换脑子
进度头跟着变：现在用哪个模型、上下文占了多少、Goal 跑到第几轮
图、语音、文件收得进、发得出
重启回来还在刚才那条会话，不用重新自我介绍

一句话：把 dsh 装进飞书，而不是把飞书接到一个网页旁边。

卖点

完全不依赖 Web

不装 dsh-web-app，不开 3080，不跑 headless 一次性进程。一个 dsh --profile feishu 常驻，飞书 WebSocket 进、飞书卡片出。关掉网页，产品还在。

斜杠命令就在聊天框

/stop /clear /model /status /compact /resume /rename /goal /bye

不支持的 / 会直接告诉你本期有哪些。完整表见 docs/handbook.md。

会变的进度头

不是「正在思考…」四个字一直挂着。进度卡片原地改：

[dsf-23% G1/3] Using bash...(2)

模型别名、上下文占比、Goal 轮次（快到上限会标 near）。跑完变 Done，失败变 Failed，正文另发一条。

Goal：说一个目标，让它自己跑

/goal 把这个目录的 README 补上安装步骤，最多跑 3 轮

Agent 自己多轮推进。重启后不会偷偷接着烧，要显式 /goal resume。

一进程多 Bot

同一个 dsh 进程可以挂 2～3 个独立飞书应用。会话按 (appId, chat_id) 分开。历史可以互相 /resume，同一条 session 不能被两个窗口同时握住。

上下文会自己收

新用户轮开始前，更早的 read / glob / grep / bash 大结果收成占位。glob 自动滤掉 node_modules / .git / .pnpm。可选：把生图 / 多步浏览类 MCP 藏到 enable_mcp 之后。

多模态按飞书的方式走

图 / 语音 / 文件落到本地 inbox，把路径写给模型。不把图片塞进 session image block。回图、回文件、回语音走 send_file。

五分钟跑起来

需要 Node 22 和已安装的 dsh（npm i -g @deepseek-ai/dsh）。

飞书开放平台建一个企业自建应用，事件订阅用长连接（WebSocket），不要 webhook。步骤见 docs/feishu-setup.md。

从 GitHub 装到独立 profile

dsh plugin --profile feishu add github:shoxiy-danny/dsh-feishu

建议钉扫描过的 commit，和市场页证据对齐：

dsh plugin --profile feishu add github:shoxiy-danny/dsh-feishu#<sha>

然后把 App ID / Secret 和 DEEPSEEK_API_KEY 放进环境，再：

dsh --profile feishu

从本仓脚本装（开发 / 自托管）

cp .env.example .env
# 填 FEISHU_APP_ID / FEISHU_APP_SECRET / DEEPSEEK_API_KEY

chmod +x scripts/*.sh scripts/dsh-feishu-cli
./scripts/setup.sh
./scripts/start.sh

启动后到这个应用的单聊打一句。应先收到表情，再看到进度头，然后是正文。

工作目录默认家目录；附件落 workspace/inbox/。第二个 bot 再填 FEISHU_APP_ID_2 / FEISHU_APP_SECRET_2。

和「再包一层飞书 Bot」的差别

多数方案是：外面一个 webhook 进程，里面再 spawn 一次 CLI。这条路对不上常驻、补话、进度、/stop、重启续聊。

这是 dsh 的 Cordis 插件。Agent 的 create / followup / steer / cancel / resume 都在同一进程里。飞书是面，不是壳外面再套的壳。

适合谁

已经在用飞书办公，不想再开一个 Agent 网页
想把 dsh 做成 7×24 云端同事，而不是本地 TUI
需要同一套 Agent、多个飞书身份
想要 Goal 这种「交代完让它自己盯」的能力，但中断和续跑必须自己说了算

不适合：还想保留官方 Web 控制台、需要人工点审批弹窗、或把飞书只当通知渠道的人。

文档

文件	内容
docs/handbook.md	斜杠命令、Goal、模型、多 bot、MCP、假 bot
docs/feishu-setup.md	飞书应用、权限、长连接
docs/publish.md	作者：GitHub topic 和市场收录
examples/	模型表、profile overlay 样例

默认模型只有 DeepSeek 官方 dsf / dsp。加自己的供应商见手册「模型」一节。

需要 DeepSeek Harness。发现页：dsh-plugin.market。

English

Feishu / Lark as the only UI for DeepSeek Harness. No web app. No TUI. No approval popup. Models, slash commands, progress, Goal, and sessions all live in chat.

The official dsh surface is a website. This bundle replaces it with one long-lived process and a Feishu WebSocket.

Why Feishu, not another web UI

Coding agents are already strong. What's missing is an entry you already open every day.

Feishu is not a side-channel for notifications. It is the product:

Slash commands in chat, same effect as a terminal
Hot-swap models; the next turn uses the new brain
A live progress header: which model, how full the context is, which Goal round
Images, voice, and files in and out
After a restart you are still in the same session

In one line: put dsh inside Feishu, don't bolt Feishu onto a webpage.

Features

No web dependency

No dsh-web-app, no port 3080, no one-shot headless runner. dsh --profile feishu stays up. Feishu WebSocket in, Feishu card out. Close the browser; the product is still there.

Slash commands in the composer

/stop /clear /model /status /compact /resume /rename /goal /bye

Unknown / replies with the supported list. Full table: docs/handbook.md.

A progress header that actually changes

Not a frozen "Thinking…". The card edits in place:

[dsf-23% G1/3] Using bash...(2)

Model alias, context %, Goal round (near when the cap is close). Done / Failed on the card; the reply is a separate message.

Goal: name an objective, let it run

/goal add an install section to this README, max 3 rounds

The agent continues across turns. After a process restart it will not keep burning tokens until you say /goal resume.

Several bots, one process

The same dsh process can hold 2–3 Feishu apps. Sessions are keyed by (appId, chat_id). History can /resume across bots; one session cannot be held by two windows at once.

Context trims itself

Before a new user turn, older bulky read / glob / grep / bash results collapse to placeholders. glob drops node_modules / .git / .pnpm. Optional: hide vision / multi-step browser MCP until enable_mcp.

Multimodal the Feishu way

Inbound images / voice / files land in a local inbox; the model gets a path. Images are not stuffed into session image blocks. Replies go through send_file.

Five minutes

Need Node 22 and dsh (npm i -g @deepseek-ai/dsh).

Create a Feishu enterprise self-built app. Subscribe to events over WebSocket, not webhook. Details: docs/feishu-setup.md.

Install from GitHub

dsh plugin --profile feishu add github:shoxiy-danny/dsh-feishu

Pin a scanned commit if you want the same evidence the market page shows:

dsh plugin --profile feishu add github:shoxiy-danny/dsh-feishu#<sha>

Put FEISHU_APP_ID / FEISHU_APP_SECRET / DEEPSEEK_API_KEY in the environment, then:

dsh --profile feishu

Install from this repo (dev / self-host)

cp .env.example .env
# fill FEISHU_APP_ID / FEISHU_APP_SECRET / DEEPSEEK_API_KEY

chmod +x scripts/*.sh scripts/dsh-feishu-cli
./scripts/setup.sh
./scripts/start.sh

Then send one message in a p2p chat with that app. You should get a reaction, a progress header, then the reply.

Default cwd is the home directory; attachments land in workspace/inbox/. A second bot is FEISHU_APP_ID_2 / FEISHU_APP_SECRET_2.

Not "a webhook that shells out to a CLI"

Most Feishu bots spawn a CLI per message. That cannot do a long-lived agent, mid-turn followups, live progress, /stop, or resume after restart.

This is a Cordis plugin for dsh. create / followup / steer / cancel / resume stay in one process. Feishu is the surface, not a wrapper around a wrapper.

Who this is for

Already lives in Feishu, does not want another Agent webpage
Wants dsh as a 24/7 coworker, not a local TUI
Needs one agent, several Feishu identities
Wants Goal ("leave it running") but pause / resume must be explicit

Not for: people who still want the official web console, a human approval popup, or Feishu-as-notifications-only.

Docs

File	What
docs/handbook.md	Commands, Goal, models, multi-bot, MCP, fake CLI bot
docs/feishu-setup.md	Feishu app, scopes, long connection
docs/publish.md	Author: GitHub topics and market listing
examples/	Model table and profile overlay samples

Ships DeepSeek official aliases only: dsf / dsp. Extra providers: see the handbook.

Requires DeepSeek Harness. Discovery: dsh-plugin.market.

MIT. Issues and PRs welcome.

dsh-feishu

中文 · English

中文

DeepSeek Harness 的飞书前端。 没有网页，没有 TUI，没有审批弹窗。模型、斜杠命令、进度、Goal、会话，全部发生在飞书里。

dsh 官方默认面是 Web。这套插件把它换成一条长连接：进程常驻，飞书就是整个产品。

为什么是飞书，不是再做一个网页

写代码的 Agent 已经很强。缺的是一个你每天都在用的入口。

飞书不是旁路通知，是唯一交互面：

斜杠命令在飞书里敲，效果跟终端一样
模型热切换当场生效，下一轮就换脑子
进度头跟着变：现在用哪个模型、上下文占了多少、Goal 跑到第几轮
图、语音、文件收得进、发得出
重启回来还在刚才那条会话，不用重新自我介绍

一句话：把 dsh 装进飞书，而不是把飞书接到一个网页旁边。

卖点

完全不依赖 Web

不装 dsh-web-app，不开 3080，不跑 headless 一次性进程。一个 dsh --profile feishu 常驻，飞书 WebSocket 进、飞书卡片出。关掉网页，产品还在。

斜杠命令就在聊天框

/stop /clear /model /status /compact /resume /rename /goal /bye

不支持的 / 会直接告诉你本期有哪些。完整表见 docs/handbook.md。

会变的进度头

不是「正在思考…」四个字一直挂着。进度卡片原地改：

[dsf-23% G1/3] Using bash...(2)

模型别名、上下文占比、Goal 轮次（快到上限会标 near）。跑完变 Done，失败变 Failed，正文另发一条。

Goal：说一个目标，让它自己跑

/goal 把这个目录的 README 补上安装步骤，最多跑 3 轮

Agent 自己多轮推进。重启后不会偷偷接着烧，要显式 /goal resume。

一进程多 Bot

同一个 dsh 进程可以挂 2～3 个独立飞书应用。会话按 (appId, chat_id) 分开。历史可以互相 /resume，同一条 session 不能被两个窗口同时握住。

上下文会自己收

多模态按飞书的方式走

图 / 语音 / 文件落到本地 inbox，把路径写给模型。不把图片塞进 session image block。回图、回文件、回语音走 send_file。

五分钟跑起来

需要 Node 22 和已安装的 dsh（npm i -g @deepseek-ai/dsh）。

飞书开放平台建一个企业自建应用，事件订阅用长连接（WebSocket），不要 webhook。步骤见 docs/feishu-setup.md。

从 GitHub 装到独立 profile

dsh plugin --profile feishu add github:shoxiy-danny/dsh-feishu

建议钉扫描过的 commit，和市场页证据对齐：

dsh plugin --profile feishu add github:shoxiy-danny/dsh-feishu#<sha>

然后把 App ID / Secret 和 DEEPSEEK_API_KEY 放进环境，再：

dsh --profile feishu

从本仓脚本装（开发 / 自托管）

cp .env.example .env
# 填 FEISHU_APP_ID / FEISHU_APP_SECRET / DEEPSEEK_API_KEY

chmod +x scripts/*.sh scripts/dsh-feishu-cli
./scripts/setup.sh
./scripts/start.sh

启动后到这个应用的单聊打一句。应先收到表情，再看到进度头，然后是正文。

工作目录默认家目录；附件落 workspace/inbox/。第二个 bot 再填 FEISHU_APP_ID_2 / FEISHU_APP_SECRET_2。

和「再包一层飞书 Bot」的差别

多数方案是：外面一个 webhook 进程，里面再 spawn 一次 CLI。这条路对不上常驻、补话、进度、/stop、重启续聊。

这是 dsh 的 Cordis 插件。Agent 的 create / followup / steer / cancel / resume 都在同一进程里。飞书是面，不是壳外面再套的壳。

适合谁

已经在用飞书办公，不想再开一个 Agent 网页
想把 dsh 做成 7×24 云端同事，而不是本地 TUI
需要同一套 Agent、多个飞书身份
想要 Goal 这种「交代完让它自己盯」的能力，但中断和续跑必须自己说了算

不适合：还想保留官方 Web 控制台、需要人工点审批弹窗、或把飞书只当通知渠道的人。

文档

文件	内容
docs/handbook.md	斜杠命令、Goal、模型、多 bot、MCP、假 bot
docs/feishu-setup.md	飞书应用、权限、长连接
docs/publish.md	作者：GitHub topic 和市场收录
examples/	模型表、profile overlay 样例

默认模型只有 DeepSeek 官方 dsf / dsp。加自己的供应商见手册「模型」一节。

需要 DeepSeek Harness。发现页：dsh-plugin.market。

English

Feishu / Lark as the only UI for DeepSeek Harness. No web app. No TUI. No approval popup. Models, slash commands, progress, Goal, and sessions all live in chat.

The official dsh surface is a website. This bundle replaces it with one long-lived process and a Feishu WebSocket.

Why Feishu, not another web UI

Coding agents are already strong. What's missing is an entry you already open every day.

Feishu is not a side-channel for notifications. It is the product:

Slash commands in chat, same effect as a terminal
Hot-swap models; the next turn uses the new brain
A live progress header: which model, how full the context is, which Goal round
Images, voice, and files in and out
After a restart you are still in the same session

In one line: put dsh inside Feishu, don't bolt Feishu onto a webpage.

Features

No web dependency

No dsh-web-app, no port 3080, no one-shot headless runner. dsh --profile feishu stays up. Feishu WebSocket in, Feishu card out. Close the browser; the product is still there.

Slash commands in the composer

/stop /clear /model /status /compact /resume /rename /goal /bye

Unknown / replies with the supported list. Full table: docs/handbook.md.

A progress header that actually changes

Not a frozen "Thinking…". The card edits in place:

[dsf-23% G1/3] Using bash...(2)

Model alias, context %, Goal round (near when the cap is close). Done / Failed on the card; the reply is a separate message.

Goal: name an objective, let it run

/goal add an install section to this README, max 3 rounds

The agent continues across turns. After a process restart it will not keep burning tokens until you say /goal resume.

Several bots, one process

The same dsh process can hold 2–3 Feishu apps. Sessions are keyed by (appId, chat_id). History can /resume across bots; one session cannot be held by two windows at once.

Context trims itself

Multimodal the Feishu way

Inbound images / voice / files land in a local inbox; the model gets a path. Images are not stuffed into session image blocks. Replies go through send_file.

Five minutes

Need Node 22 and dsh (npm i -g @deepseek-ai/dsh).

Create a Feishu enterprise self-built app. Subscribe to events over WebSocket, not webhook. Details: docs/feishu-setup.md.

Install from GitHub

dsh plugin --profile feishu add github:shoxiy-danny/dsh-feishu

Pin a scanned commit if you want the same evidence the market page shows:

dsh plugin --profile feishu add github:shoxiy-danny/dsh-feishu#<sha>

Put FEISHU_APP_ID / FEISHU_APP_SECRET / DEEPSEEK_API_KEY in the environment, then:

dsh --profile feishu

Install from this repo (dev / self-host)

cp .env.example .env
# fill FEISHU_APP_ID / FEISHU_APP_SECRET / DEEPSEEK_API_KEY

chmod +x scripts/*.sh scripts/dsh-feishu-cli
./scripts/setup.sh
./scripts/start.sh

Then send one message in a p2p chat with that app. You should get a reaction, a progress header, then the reply.

Default cwd is the home directory; attachments land in workspace/inbox/. A second bot is FEISHU_APP_ID_2 / FEISHU_APP_SECRET_2.

Not "a webhook that shells out to a CLI"

Most Feishu bots spawn a CLI per message. That cannot do a long-lived agent, mid-turn followups, live progress, /stop, or resume after restart.

This is a Cordis plugin for dsh. create / followup / steer / cancel / resume stay in one process. Feishu is the surface, not a wrapper around a wrapper.

Who this is for

Already lives in Feishu, does not want another Agent webpage
Wants dsh as a 24/7 coworker, not a local TUI
Needs one agent, several Feishu identities
Wants Goal ("leave it running") but pause / resume must be explicit

Not for: people who still want the official web console, a human approval popup, or Feishu-as-notifications-only.

Docs

File	What
docs/handbook.md	Commands, Goal, models, multi-bot, MCP, fake CLI bot
docs/feishu-setup.md	Feishu app, scopes, long connection
docs/publish.md	Author: GitHub topics and market listing
examples/	Model table and profile overlay samples

Ships DeepSeek official aliases only: dsf / dsp. Extra providers: see the handbook.

Requires DeepSeek Harness. Discovery: dsh-plugin.market.

MIT. Issues and PRs welcome.

dsh-feishu

dsh-feishu

中文

为什么是飞书，不是再做一个网页

卖点

完全不依赖 Web

斜杠命令就在聊天框

会变的进度头

Goal：说一个目标，让它自己跑

一进程多 Bot

上下文会自己收

多模态按飞书的方式走

五分钟跑起来

从 GitHub 装到独立 profile

从本仓脚本装（开发 / 自托管）

和「再包一层飞书 Bot」的差别

适合谁

文档

English

Why Feishu, not another web UI

Features

No web dependency

Slash commands in the composer

A progress header that actually changes

Goal: name an objective, let it run

Several bots, one process

Context trims itself

Multimodal the Feishu way

Five minutes

Install from GitHub

Install from this repo (dev / self-host)

Not "a webhook that shells out to a CLI"

Who this is for

Docs

Community comments

dsh-feishu

dsh-feishu

中文

为什么是飞书，不是再做一个网页

卖点

完全不依赖 Web

斜杠命令就在聊天框

会变的进度头

Goal：说一个目标，让它自己跑

一进程多 Bot

上下文会自己收

多模态按飞书的方式走

五分钟跑起来

从 GitHub 装到独立 profile

从本仓脚本装（开发 / 自托管）

和「再包一层飞书 Bot」的差别

适合谁

文档

English

Why Feishu, not another web UI

Features

No web dependency

Slash commands in the composer

A progress header that actually changes

Goal: name an objective, let it run

Several bots, one process

Context trims itself

Multimodal the Feishu way

Five minutes

Install from GitHub

Install from this repo (dev / self-host)

Not "a webhook that shells out to a CLI"

Who this is for

Docs

Community comments