プラグイン

現在の dsh セッションの実行中の指示チェーンとプロジェクトスキルを検査・管理できます。その場での作成・編集にも対応し、表示されている内容がそのままモデルに渡されます。

DeepSeek Harness 向けのリクエストコンテキストプロファイラー。モデルへのリクエスト間で何が変化したか、それに伴いキャッシュの再利用がどう変わったかを確認できます。

DeepSee: DeepSeek Harness にビジョン機能、マルチモデルルーティング、納品前の Gemini による視覚的セルフチェックを提供します。

テキスト専用の DeepSeek Harness(dsh)モデルにビジョン機能をプラグインとして追加します。低価格/無料の VLM プリセットを内蔵した `vision` ツール、複数画像の一括解析、貼り付けによる画像入力、ホットリロード対応の Web 設定ページを提供します。

ブラウザの Web Speech API による音声入力。サーバー・API キー・モデルのダウンロードは不要です(Edge=Azure、Chrome=Google の音声認識を使用)。

DeepSeek Harness 向けの再利用可能な Markdown プロンプトプロファイル。ターンごとのモデル選択、引数の置換、状態の復元に対応します。

ターン、モデルのステップ、ツール呼び出しを yiTrace へエクスポートするテレメトリバックエンド。

モデル向けのセッションをまたぐメモリ機能。過去のすべてのセッションを全文検索し(ctx.sessionQuery 経由の SQLite FTS5)、最も一致度の高い抜粋を現在のコンテキストに呼び戻します。

A floating VRM avatar with tool-driven expressions and VRMA animations, plus model and motion asset management.

Simulated FX trading arena played against an algorithmic whale-chan opponent: a live canvas candlestick chart over five instruments (EUR/TOKEN, GBP/TOKEN, TOKEN/JPY, gold, and Whale/TOKEN, her own coin she never shorts), selectable leverage from 10x to 500x, and two modes — a timed mode with 90s to 500s rounds settled on equity, and an untimed goal mode that starts with $300,000 and ends when someone first reaches $20M or goes into debt (equity below zero, no maintenance margin). The match runs in a module-level engine, so switching to a conversation does not stop it or restart it. Margin and spread are modeled, prices are a local random walk with occasional news shocks, and nothing touches the network or real money.

Map a remote server directory over SSH and SFTP into a controlled local cache workspace, so the model, agent, session and memory keep running on the local machine.

Pairs a phone console with the current DSH Desktop conversation. Running /phone shows a one-time QR code; after pairing the phone can send messages, approve tool calls, switch models and stop the task, while the desktop stays the only execution host. Requires AgentBridge running locally.

Redacts PII in tool-result text before it is logged (error-result metadata is not rewritten), denies read_image and binary/document reads outside allowPaths, provides local privacy_extract, privacy_ocr, privacy_scan and privacy_rehydrate tools (PDF/OCR needs a separately installed local extractor configured through extractCli), and enforces a fail-closed llm/stream egress guard before DSH dispatches any model request.

API key switcher for DeepSeek Harness: save multiple API keys in Settings and switch the active one; the next model request uses it, with no DSH restart.

Records a session as a run, replays it in an isolated workspace copy with another model or prompt, diffs the two trajectories side by side, aggregates runs into a task × variant matrix, and gates regressions with task assertions and a CI exit code.

Compare Git/SVN versions across repositories and export Chinese Markdown release reports with optional approved model analysis.

Keeps a model request alive through a dropped connection: transport failures retry until the network returns while real errors still fail fast, and a turn killed by a process crash offers a one-click, state-checking continuation.

Left-sidebar code review panel for DeepSeek Harness. Reviews local diffs, commits, or GitHub pull requests with the configured model using the Codex review rubric and P0-P3 findings. GitHub PR review requires a logged-in `gh` CLI.

Talk to DeepSeek Harness through Siri and iPhone Shortcuts: a bearer-key POST /siri endpoint on your computer answers each spoken request for Siri to read aloud, with a settings page (key, port, workspace, model, Agent Preset, rate limit, allowed addresses), optional DS Bot routing, and a ready-made Shortcut.

Hold Alt (default; rebindable in the plugin page) to dictate and insert a cleaned-up transcript, plus a sparkle button that rewrites the composer draft into a clearer prompt. Dictation requires the Harness Voice Input plugin with its SenseVoice model enabled.

Video study toolkit for DeepSeek Harness: fetch a public Bilibili single part or YouTube video, sample every 3 seconds plus change-triggered frames, transcribe locally with faster-whisper, page through section grids, and render a 16:9 study deck. Windows only, and needs the DSH pwsh tool plus Python 3.12 x64 with Pillow and FFmpeg/FFprobe; fetching adds yt-dlp, transcription needs faster-whisper with your own local model files, and deck output needs python-pptx.

Fusion 360 for DSH: bridges the MCP server Fusion ships (read, execute, update, electronics) with every upstream schema kept byte-for-byte while the always-on tool surface drops 70 percent, adds an editable endpoint under Settings, and wraps a local hf.py modelling loop as nine hf_* tools (needs Python 3.10+ and a fusion-harness workspace).

Layered memory for AI agents: five verbs (wake/work/pit/tomb/close), core layer read every time, archive on demand. Plain markdown, four pages created on first run; survives restarts, model swaps and plugin removal.

Restarts the DSH desktop host on macOS when a change needs it, then hands the conversation back to the same session with a note the model writes about why it restarted and what comes next.