Plugins

Scheduled tasks: run a prompt on cron schedules, each run opens a new agent session to do the work, with workspace binding, manual run, session auto-naming, and a fullscreen task management page.

Records every agent write, edit and shell `rm` into a shadow git repository, then reviews them hunk by hunk in VS Code — accept or revert each one. Needs the companion dsh-review-vscode extension; the panel only appears inside VS Code's webview.

Cross-session memory with zero-token deterministic recall (bigram-Jaccard fused with FTS5 BM25 via RRF, Chinese-aware tokenization), reinforcement that strengthens repeated observations, and procedures crystallized into SKILL.md files refined in place through an active-stale-archived lifecycle with reversible archive and rollback. Ingest and disposal each expose a user-chosen autonomy tier; conflicts and high-importance writes always go to human review, and nothing is ever auto-deleted. Requires Node 22.5+.

A self-healing layer for tool calls: missing params, wrong fields and out-of-range paths are fixed before they reach the model. Measured visible error rate fell from 7.95% to 2.20%.

Inline +N −M badges on edit/write tool rows and a collapsible per-turn file-change summary card with aligned diffs, review and undo — covers native edit/write, str_replace_editor and Code Dispatch sub-calls, no git required.

Automatically and intelligently routes each delegated task to the best configured model using aliases, capability tags, and usage guidance—keeping routine work fast and affordable while sending complex work to stronger specialists. Configure your model team in the Web UI or guided setup, run specialists in parallel with reliable result joining, and see the chosen model at a glance in subagent views and inline in Better Sidebar's Tasks tree.

Voice input for the web UI: dictation chunked by pauses and voice messages, each with its own provider fallback chain (Deepgram, Groq, HuggingFace, local whisper.cpp, or an OpenAI-compatible endpoint).

Official Remotion skill ported to DSH: programmatic video with React (animation/audio/captions/3D/charts/fonts, 38 rule files), plus a bundled-resource health check.

Global Chinese mode: a compact pill switch beside the composer that injects a language requirement into any preset's system prompt (complete personas excepted); reply, thinking and tool output each toggle between Chinese and English, and anchored presets are supported.

Multi-machine sync: keeps multiple dsh replicas consistent through one private GitCode repository — skills, sessions, settings and plugin manifests each with an independent switch; changes go through branch, PR and merge, remote-only additions are pulled back before every push so nothing is deleted, and an AI smart-align step semantically merges files both sides changed (plus one-click conflict resolution); private repos enforced, pull never overwrites local edits.

Six-player no-limit Texas Hold'em for the dsh Web GUI. One human plays against five LLM agents that each see only their own hole cards. Adds a dedicated tab plus a draggable mini-window over the conversation area, per-browser sessions with CSRF protection, pot-relative raise sizing, three rebuys before elimination, and showdown reveals with board-card deal animations.

Live security beacon for DSH that follows tool activity. Click to view call statistics, risk provenance chains and blocked actions, with read-only task restrictions and tool metadata change detection.

Learning Mode agent preset: a coding agent that teaches while coding — concrete scenario-grounded explanations, Socratic guidance, and TODO(你) practice blanks, modeled on Claude Code's Learning output style; installable via dsh plugin add (dsh-learning-mode on npm).

Dead-code detection for AI-generated codebases: graphlint builds a dependency graph, finds code unreachable from any entry point to enable codebase cleanup and functional‑effectiveness understanding, and exposes graphlint_query / graphlint_build / graphlint_config tools plus a graphlint skill to DSH agents.

Literary DSH plugin: one session is one book. Cover-card /new opens one of 19 realms, a prose-only 诸天万界DSH preset keeps the model writing fiction, keyword worldbook entries ground each turn, and /export-story typesets the session into a Markdown novel.

Terminal-native interactive TUI for DeepSeek Harness (dsh) — built with Ink, React for terminals

Solarized theme plugin: the classic low-contrast, eye-friendly dual palette system — Ethan Schoonover's Solarized plus Selenized (its refined take), dark and light each, built for long sessions with canonical syntax colors.

Let dsh and Claude Code sessions message each other directly; comes with a clickable peer list card (sort/search/send/refresh).

A 60-frame theme-aware monochrome whale-dive animation with reactive water for the DSH Web turn status: embedded assets, seamless closed-loop playback, a reduced-motion PNG fallback, and lifecycle-clean removal.

Custom modes / assistant manager: give each mode its own system prompt in the settings page — a custom prompt that is a plain file the agent loop re-reads before every model step, so an edit applies to the running session on its next step — and start it from the official standard / PTC / minimal / Cordis composition, toggling the plugin rows it mounts one by one (a custom plugin composition per mode), renaming it as you like. Several assistants coexist with their own prompts, so one machine can keep a coding assistant, a chat companion and a roleplay (RP) persona side by side — a multi-agent setup you pick from per new session. Rows left untouched stay byte-identical to the shipped ones, platform conditions included.

Web appearance suite with a 7-theme gallery (each a full 122-token --dsw-* mapping) plus one-click switching persisted in settings, Material You wallpaper color extraction that generates a whole dynamic palette, and a global corner-radius control. Themes ride on the official look system as override layers instead of fighting it, so the official light/dark/auto row stays valid.

Custom modes / assistant manager: give each mode its own system prompt in the settings page — a custom prompt that is a plain file the agent loop re-reads before every model step, so an edit applies to the running session on its next step — and start it from the official standard / PTC / minimal / Cordis composition, toggling the plugin rows it mounts one by one (a custom plugin composition per mode), renaming it as you like. Several assistants coexist with their own prompts, so one machine can keep a coding assistant, a chat companion and a roleplay (RP) persona side by side — a multi-agent setup you pick from per new session. Rows left untouched stay byte-identical to the shipped ones, platform conditions included.

Query Chinese Minecraft Wiki (MediaWiki API) and convert raw wiki text into clean AI-readable output — search, page intro, full-page Markdown, and random entry; templates, references, images and cross-language links are stripped before reaching the model.

Per-step model router: named presets with ordered image/keyword-group rules route each agent step across every mounted provider and model, adding collaboration flows for image transcription, an image guard, a multi-plan quota panel, and per-step decision observability.