MIT native DSH plugin for a persistent companion: memory, diary, mood, routines, optional WeChat/QQ/browser/desktop — all gated in the World Console.
DSH integration
Native runtime
Author-claimed
Safety audit
Unaudited
Last verified
2026-11-20
License
MIT
01What can it help you accomplish?
Keep a DSH companion that remembers chats, writes her own diary and follows a daily rhythm
Recent turns stay in context; older lines are retrieved via `memory_find` / `memory_read`; she maintains diary, mood and wiki pages under her room
People who want a long-running persona inside DSH instead of a stateless chat window
Talk to the same companion on WeChat or QQ with one shared memory
Private and group chats route into the home session; unread summaries appear in her context without forcing replies
People on Windows with the WeChat desktop build the adapter supports; for QQ, anyone who separately installs SnowLuma OneBot—README does not limit QQ to Windows
Optionally let her browse with her own Edge or operate the desktop after you enable switches
Browser or desktop flows save as JSON procedures under the world directory; foreground mode uses real mouse and keyboard, stop with double Esc
Power users who accept high-touch desktop automation under explicit console toggles
02How to install into DeepSeek Harness
Prerequisites
- DSH desktop app — the World Console and life line depend on it
- Before first install or upgrade, read「默认就会发生的事」in NOTES.md as the README suggests
Installation steps
- 01
In your DSH profile directory, run `node <DSH 自带 pnpm 的入口文件> add dsh-world@latest --ignore-scripts --registry=https://registry.npmjs.org/`
- 02
Register the bundle in the profile per INSTALL.md, then restart DSH
- 03
Open「世界控制台」in DSH settings and fill persona under「人设」
- 04
Press「开始」in the console so the plugin builds her room and home
Verify the integration
- Pressing「开始」creates her room and home and wakes her — that is the README success signal
03DSH integration and capability boundaries
Cordis bundle in the DSH profile; uses only DSH open extension APIs, with a World Console in DSH settings.
Layered memory and mood
Chat turns in the world workspace, diary folder and wiki pages she writes→`memory_find`, `memory_read`, `memory_image`, nightly「我是谁」and mood snapshots in context
Writes archives, room, diary and channel files under each world directory (see README section「她的东西放在哪」)Home life line and World Console
Persona, alarms, quotas and start/stop from DSH settings→A home session, scheduled review, spend tracking and module switches in one console
Updates `world.json`, `life.json`, persona files; can replace profile compaction with `dsh-world/compaction` when configuredWeChat and QQ channels
WeChat Windows adapter or SnowLuma OneBot WebSocket config in `cordis.patch.yml`→`wechat_send` / QQ outbound messages sharing the same home memory
Stores WeChat tokens in `wechat.json`; QQ settings in `qq-native.json`; connects outbound to OneBot when enabledHer own browser
Console「她的浏览器」switch after allowing `computer` in config→Separate Edge profile; learned `page_*` flows saved as JSON procedures
Creates `browser/` and `procedures/` under the world directoryDesktop control
Console「她用电脑」— requires her browser on Windows→Background single-window or foreground full-screen control; procedures stored with `where: desktop` / `screen`
Captures window or screen images and injects clicks/keys into target windows when enabled
04Who is it for? When not to use it?
Good for
- People who want a long-running persona inside DSH instead of a stateless chat window
- People on Windows with the WeChat desktop build the adapter supports; for QQ, anyone who separately installs SnowLuma OneBot—README does not limit QQ to Windows
- Power users who accept high-touch desktop automation under explicit console toggles
Not for
- The World Console and life line need the DSH desktop app; README does not describe a headless-only path.
- WeChat, QQ, her browser and desktop control are allowed in config but stay off until you flip switches in World Console → 接入.
- Foreground desktop mode uses the real mouse and keyboard; each switch asks for admin elevation and expires after reboot.
- WeChat needs a matching desktop version, Python per adapter docs, and carries account risk per the README.
- Repo is MIT; the WeChat native send engine uses a separate non-commercial license in that module.
05Compatibility, maintenance and safety notes
- The World Console and life line need the DSH desktop app; README does not describe a headless-only path.
- WeChat, QQ, her browser and desktop control are allowed in config but stay off until you flip switches in World Console → 接入.
- Foreground desktop mode uses the real mouse and keyboard; each switch asks for admin elevation and expires after reboot.
- WeChat needs a matching desktop version, Python per adapter docs, and carries account risk per the README.
- Repo is MIT; the WeChat native send engine uses a separate non-commercial license in that module.
- README header badges require DSH ≥ 0.2.1 and Node ≥ 22.16 (shields.io badge labels in the repo README).
- Default `updateCheck: true` asks npm once per day whether a new version exists; set it to false to skip that network call.
MIT · last push 2026-11-18 · release v0.8.2
06Frequently asked questions
Why does the README insist on DSH desktop?
The World Console and the life line live in DSH settings on the desktop app. The quick-start section opens with「需要 DSH 桌面版」and ties those features to it.
Are WeChat, QQ, browser and desktop on after install?
No. The feature table marks them「默认关」, and config flags like `wechat` / `computer` / `desktop` only allow modules — you must open them under World Console → 接入.
What is the manual install path?
Run the bundled pnpm add command in your profile, register the bundle per INSTALL.md, restart DSH, set persona in the console, then press Start. INSTALL.md also documents tarball updates with backup.
Where does world data live?
README「她的东西放在哪」lists files under each world directory; default DSH_HOME is the `.dsh` folder in your user profile. Back up the whole tree before upgrades.
What should I read before first install?
README points to NOTES.md「默认就会发生的事」and the model hand-off line about not enabling features you did not approve. It also suggests reading upgrade notes in CHANGELOG when updating.
07Related DSH workflows
dsh-univer-office
by dream-num
Give DeepSeek Harness a real office environment. Univer Office Plugin brings spreadsheets, docs, slides, canvases, relational tables, and more into one runtime — with connected data, validation, versioned changes, and isolated worktrees for multi-agent collaboration.
clearai-dsh
by clearailhc
Brings an evidence-checked epistemic loop to DSH: research sessions settle into a browsable domain ontology with evidence chains and support levels; contradictory readings are surfaced for human adjudication.
dsh-fish-memory
by yuuyuko-uu
Stores time-stamped, decaying per-session memories, recalls relevant items, and optionally distills imported local archives through the configured DSH model.
dsh-trading
by zhu1090093659
Agent-native trading terminal built on DeepSeek Harness. Crypto, US, CN and HK in one three-column GUI, 19+ hot-swappable connectors, dry-run by default with human approval on every live order. BYOK, no data redistribution.
08Data and sources
给 DSH 里的陪伴角色一个会记得、会生活的世界。
插件只用 DSH 原本开放的扩展接口,不改宿主的一行代码。
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-11-20. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.
