DeepSeek Harness Plugin Hub

Publish and manage complete Harness Profiles. Discover Plugins for your next setup.

Explore

PluginsPresetsDocsNews

Community

Publish a pluginContactReport an issue

Resources

Plugin Hub on GitHubDeepSeek HarnessSystem statusPrivacy notice
© 2026 DeepSeek Harness Plugin HubPowered byPaxTech

Independent and unofficial. Not affiliated with, authorized by, or endorsed by DeepSeek.

Adhd Session Mode — DSH Plugin for DeepSeek Harness
← Plugins
A

dsh-adhd-session-mode

Adhd Session Mode

DeepSeek Harness plugin converted from ayghri/i-have-adhd: ADHD-friendly output shaping that takes effect in a single session, never globally.

The plugin will be installed here. Keep web if you are unsure.

npx -y @deepseek-ai/dsh plugin --profile web add github:StarterMonk/dsh-adhd-session-mode#43429acc37b45bf68ed10a54dfe09e95409094e4
READMECompatibilityVersions

Compatibility and provenance

Adhd Session Mode is published as dsh-adhd-session-mode and currently resolves to version 1.0.0. The Hub verifies its manifest and preserves the exact installation source for reproducible installs.

DSH compatibility
*
Runtime surfaces
web
Release source
github
Registry updated
10/8/2026

Versions

1.0.0stable
10/8/2026

Related plugins

Loading related plugins…

Latest
1.0.0
DSH
*
HMR
Process restart
Tree shaking
Safe tree shaking not declared
Unpacked size
Unavailable
Files
Unavailable
Surface
web
License
MIT
Source
github
GitHub
★ 0
Weekly downloads
0
Last push
10/8/2026
View source ↗
README badge

Click the badge to copy Markdown for your README.

Do you maintain this Plugin?Claim benefit · Priority security scan

Verify the GitHub repository declared in package.json to manage this listing. After you claim it, Hub will prioritize a security scan of the current version and publish the result when it passes.

Claim this Plugin →
Report an issue
DeepSeek Harness Plugin Hub
ProfilesPluginsCategoriesNewsDocsSign inManage Profiles
ProfilesPluginsCategoriesNewsDocsSign in

Related plugins

More verified plugins in productivity-workflow.

Acp App@deepseek-ai/dsh-acp-appThe dsh ACP profile bundle: automation-only JSON-RPC stdio and process lifecycle over dsh-baseExperimental Schedule Bundle@deepseek-ai/dsh-experimental-schedule-bundleOptional bundle that adds Schedule, its task page, and time context to the Web compositionAb Ocrdsh-ab-ocr--- description: "The out-of-tree ocr tool: recognizes a scanned PDF or image page by page, merges the pages in order only after the last one, and saves one Markdown document with the heading levels the model decided." kind: "package-reference" ---Deepseek Ipptdeepseek-ipptiPolloWork PPT Studio and its curated slide templates as a native DeepSeek Harness conversation view.

README

dsh-adhd-session-mode

English · 简体中文

A DeepSeek Harness (dsh) plugin converted from the GitHub skill project ayghri/i-have-adhd. It takes effect in one session at a time — never globally.

One system-prompt section rewrites how the assistant replies — action first, numbered steps, concrete time estimates, no preamble or closers — and every session switches it on or off on its own.

Origin

The upstream project is an output-style skill for several coding assistants and ships no DeepSeek Harness code. This repository ports its ruleset to DSH and adds the harness integration the original had no equivalent for: a system-prompt section, a /i-have-adhd command, a session-log projection that holds the per-session state, and a badge below the composer.

The ruleset in skill/SKILL.md is upstream's text reproduced verbatim under its MIT licence — see THIRD-PARTY-NOTICES.md. Everything else here is original.

Why per-session

Output style is not a global preference; it depends on what the conversation is for.

Sometimes you need the model to explain something properly — to walk through a mechanism, weigh trade-offs, or answer a question where the detail is the answer. A mode that compresses every reply into a numbered action list actively gets in the way there. At other times you are executing: you know what you want, you want the next move and nothing around it, and a five-paragraph lead-in is pure cost.

A single global switch forces one of those to lose. So this plugin makes the mode a property of the session:

  • Turn it on in the sessions where you are executing.
  • Leave it off in the sessions where you are reading and thinking.
  • Both can be open at the same time, and neither remembers the other's choice.

It is not a setting you flip and forget; it is a decision you make per conversation, which is the granularity at which the decision is actually made.

Install

dsh plugin --profile <your-profile> add dsh-adhd-session-mode

From the source repository instead:

dsh plugin --profile <your-profile> add github:StarterMonk/dsh-adhd-session-mode

Then restart DeepSeek Harness. The plugin's row is a bundle patch, so the first activation needs a restart.

Usage

In the session you want shaped:

/i-have-adhd on

/i-have-adhd off turns it back off, and a bare /i-have-adhd toggles. You can also just say stop adhd mode (or normal mode) as an ordinary message — the phrase is matched against the whole message, so how do I stop adhd mode? is treated as a question, not a command.

While the mode is on for a session, a small ● ADHD ON badge sits below the composer, so the state is never a thing you have to remember.

Starting every session with it on is possible too, by overriding the row's config in your profile patch:

- id: i-have-adhd
  name: dsh-adhd-session-mode
  config:
    alwaysOn: true

How it works

PieceWhat it does
System-prompt sectionCarries the ruleset into every model step while the session's mode is on. Registered at order 10150 — after every tool description, the harness source and the web surface, immediately before the deployment persona's closing section — so a style instruction lands last.
Session projectionThe mode is derived from the session log, not from a setting. /i-have-adhd on is durable as command/run; a typed stop adhd mode is durable as user/message. Folding those two events means the mode replays correctly when a session is resumed or forked.
/i-have-adhd commandThe switch. Declaring input on the command is what lets the client route /i-have-adhd on into it instead of submitting the line to the model as text.
SkillThe full ruleset is also registered as an on-demand skill, so the text can be loaded when it is wanted without turning the mode on.
Client badgeReads the session's projected state through useProjection and renders below the composer. No harness client package is imported; styling uses --dsw-alias-* theme tokens only.

Nothing is written to $DSH_HOME or to a config file. There is no flag file to find, and there is nothing to clean up if you uninstall.

Relationship to other ADHD plugins

Several plugins in this space do different jobs, and it is worth being precise:

PluginJob
dsh-i-have-adhdShapes the assistant's replies. One system-prompt section at order 50, plus three zero-argument tools (adhd_on / adhd_off / adhd_status) and a global flag file under $DSH_HOME that restores the mode at boot.
dsh-adhd-copilotCoaches the user: task breakdown, launch rituals.
adhdgofly-dsh-extHighlights parts of speech in rendered Markdown.
this pluginShapes the assistant's replies, with the switch scoped to one session and its state stored in that session's log.

The practical difference from dsh-i-have-adhd is the scope of the switch and where the state lives. A global flag file cannot express "on here, off there"; a per-session projection can, and because the projection folds the session log rather than holding a variable, the answer survives resume and fork without any extra bookkeeping. The trade-off is on the control surface: dsh-i-have-adhd can be driven by simply saying adhd mode on because the model has tools for it, whereas this plugin asks you to type the slash command.

Both reuse the same idea from ayghri/i-have-adhd.

Limitations

  • Delegated subagents do not inherit the mode. A subagent runs in its own session, and the mode is per session, so a child agent writes in the default style. Parent and child therefore differ in register. Sharing the mode with children would mean threading the parent's state through the projection's init, which is not implemented.
  • One mode per session, not per turn. Turning it on or off affects the rest of that session.
  • The injected ruleset is an instruction, not a guarantee. It shapes the model's output; it cannot force compliance. Judge it by reading a few replies.

The ruleset

Ten rules, grouped for skimming — the full text is in skill/SKILL.md:

Shape · first line is the action · number the steps · close with one next move · park side quests State · re-say where we are · estimate in real units · surface what now works Tone · errors are facts · cap lists at five · no throat-clearing

With explicit overrides: harness rules outrank the mode; an "explain" request keeps the shape but opens up the depth; destructive actions still confirm first; and a three-turn debug spiral triggers one diagnostic question instead of more edits.

Licence

MIT. The ruleset text is reproduced verbatim from ayghri/i-have-adhd (MIT) — see THIRD-PARTY-NOTICES.md, which carries the upstream licence in full. Everything else here is original.