플러그인

Article Review page in DSH's native right sidebar: turns Markdown, Word, PowerPoint and PDF articles into sentence-addressable text, lets you comment on any sentence, and hands the open review to the agent as a composer draft. Six agent tools let the model read every comment, revise the article (in place for Markdown, as a sibling <name>.revised.md for binary formats), resolve the comments, and record recurring wording preferences into a long-term writing-style memory injected into later prompts. Extraction includes a dependency-free DOCX/PPTX reader; PDF text extraction runs locally through the host's Python + pypdf (or poppler's pdftotext) and is unavailable without one of them. No network calls, no telemetry.

Windows companion window for DSH Desktop: todos, reminders, live session messages, approval handling and quick input. The standalone window reuses the Electron binary at %USERPROFILE%\.dsh\electron\electron.exe (usually installed by dsh-pet) and fails to open when it is missing.

Save any URL (intranet GitLab raw, public GitHub, etc.) to a local file — text or binary — with a text preview and skip-on-exists.

Exposes a DSH Harness Desktop Web GUI to another machine's browser through a Cloudflare quick or named tunnel, an SSH reverse tunnel, or a custom command, and hands out a tokenised link to open it; a settings panel and CLI start, stop and print the link. Needs DSH Desktop 0.2.1-rc.2 or later and a cloudflared binary.

DSH host bundle for the ShyPie privacy-first workflow: extracts attached images and PDF/Office files locally before they enter the session log (requires poppler pdftotext/pdftoppm on PATH and a local OCR engine — Windows OCR on Windows, optionally tesseract; without them attachments are rejected, not forwarded); default filter mode keeps text local (optional local masking); blocks unmaskable media/BLOB; provides local privacy_extract, privacy_ocr, privacy_scan and privacy_rehydrate tools; denies read_image and binary/document reads outside allowPaths; and enforces a fail-closed llm/stream egress guard before DSH dispatches a model request (the default provider allowlist is the local Magpie gateway).

Audio tools for DSH: inspect audio/video files, transcode or extract the audio track from a video, trim and normalize, and transcribe speech with local whisper.cpp or an OpenAI-compatible API. Requires ffmpeg and ffprobe on PATH (not bundled); local transcription also needs a whisper-cli binary and a ggml model.

One-step local model onboarding plus a light harness for DSH. On load it probes loopback and private-network inference endpoints (llama.cpp, Ollama, vLLM, LM Studio), reads each model's context window and output budget from the backend, and registers the provider through settings, so a local model becomes selectable without hand-writing config. It then trims only that provider: masks the tool list through tools.restrict, drops named system-prompt sections and memory contexts, swaps the injected instruction block for a short one while preserving its first 80 and last 96 characters, and truncates tool-schema descriptions without touching structure or required fields. Cloud providers are returned as the same assembly object, untouched. The injected block is re-read from disk on every assembly, so editing the file takes effect on the next request with no reload. The panel in the plugin detail page is a live operations view: it shows whether the takeover is active and by which hook, how many characters the last assembly was trimmed from and to, and which providers are online (probed concurrently, with a real one-shot completion reporting latency and reply text per provider). If a local backend is not running, the panel offers a one-click launch: it probes the port first (never starting a duplicate), then starts the matching server (ollama serve, or a custom command you supply) detached and polls until the endpoint answers, clamping the wait to 3-600 seconds, reaping the child it just spawned if it never answers, and naming the paths it searched when it cannot find the binary. A stop button kills backends this plugin started (SIGTERM, then SIGKILL) and refuses to touch a server somebody else launched. When a model has just been downloaded and no backend exists yet, the panel can also scan the machine for .gguf files with hard caps (300 models, 20000 directory entries, Downloads only one level deep), start llama-server on the chosen one (skipping the .mmproj. vision projection and reusing a server that already answers), register the provider under the model id the backend reports, and run one real completion to prove it works. Before spawning anything it does a fit check that counts only the part of the file the GPU cannot hold: hostBytes = fileSize - VRAM, compared against max(total RAM - 2GB, free RAM). A file far larger than RAM is refused outright; a heavy host-side remainder is refused unless you confirm; the same remainder under mmap is only a warning, because the weights land in reclaimable page cache. A load-mode selector (auto / mmap / none / mlock / mmap+mlock) maps to llama.cpp --load-mode, and the default is whatever the engine picks. From 0.4.2 everything above is derived from the platform instead of assumed: home and data directories come from os.homedir() and the OS conventions (LOCALAPPDATA on Windows, ~/Library/Application Support on macOS, XDG_DATA_HOME elsewhere); backend binaries are looked up in the official install locations of each OS and can be pinned with OLLAMA_BIN, LLAMA_SERVER_BIN, VLLM_BIN or LMSTUDIO_BIN; model discovery walks DSH_LOCAL_MODELS_DIR, ~/models, ~/llama.cpp/models, ~/.cache/llama.cpp and the LM Studio data directory, with the WorkBuddy layout demoted to one candidate among them. VRAM is asked from nvidia-smi, then rocm-smi, then sysctl hw.memsize on Apple Silicon (flagged unified, so the same bytes are not charged twice), and any flip to a unified-memory budget is handled explicitly. Before building a command line the plugin runs llama-server --help once and keeps only the flags that build actually has, so older engines get --n-gpu-layers 99 and no KV quantization while newer ones get auto plus q8_0; anything it cannot probe is passed through rather than dropped. LM Studio is started with lms server start --port N rather than being lumped in with the engines that need a model argument, and vLLM gets its own --model example instead of llama.cpp syntax. Windows, macOS and Linux are all exercised in CI, including a windows-latest and a macos-latest job, because a cross-platform claim that has only ever run on one OS is not a claim. Two of those suites exist specifically to catch this machine's blind spots: one replays the whole suite with process.platform rewritten to each OS, and one deletes every advantage this development box has — no nvidia-smi, no rocm-smi, no sysctl, an empty PATH, a home directory with no models — and requires graceful degradation rather than a crash. That last one caught a real class of bug: several assertions hardcoded 'lms.exe', which cannot pass where the executable has no suffix. Measured on the reference machine: a 16.9GB 27.2B MoE with 14.5GB of VRAM holds a 285-691MB working set and sustains 76 tok/s after a 1000-2000 token warm-up, so file size alone is not a fit signal. From 0.4.3 the scanned models are shown as a table rather than a dropdown (one row per gguf, with its size and any fit warning), so you can see what is on disk before choosing. From 0.4.4 the delete button on the provider table removes the provider entry, not the weights file on disk. That distinction matters: what people actually want to clear out is the dead entry - one pointing at a port that never answered, one registered twice for the same port, one whose model file was deleted long ago - not a 15.8GB gguf. Deleting a config is undone by importing again; deleting a gguf is not. The removal goes through settings.mutate with an unset op on \['providers', id], because update merges layers: writing null there leaves a null provider key that the loader reads as a broken config rather than an absent entry, and only mutate/unset is implemented as Reflect.deleteProperty. The result is read back afterwards, because unset on a key that comes from a lower layer (the profile cordis.patch.yml or a home patch) does not delete it - it restores the inherited value, so the call succeeds while the entry stays. When the entry is still there the panel says so and points at the file to edit instead of reporting success, and a DSH whose settings service predates mutate() is told plainly rather than silently ignored. Deleting takes two clicks and clearing the confirmation is as easy as clicking another row; on success the online and baseURL caches are cleared and the tool surface is recomputed, since the provider list changed. The gguf itself is never touched by this plugin - the scanned-model table is display and selection only.VRAM reporting is driver-dependent. From 0.4.5 model discovery also reads Ollama's own storage: its blobs are named sha256-… with no extension, so extension matching could never list them. Weights are now accepted by extension OR by GGUF magic header (only dot-free files of at least 8MB, capped at 600 header reads per scan; obviously-not-weights extensions like .json/.md/.safetensors/.part are never opened). Ollama models are resolved from their manifests: the model layer is picked by mediaType application/vnd.ollama.image.model (never adapter/mmproj layers), the human-readable name is restored from the manifest path (…/qwen3/8b becomes qwen3:8b, …/latest stays bare), tags sharing one blob merge into a single row, and the metadata kv supplies name, quantization and context length. Adopting an Ollama model does not start a duplicate llama-server over the same weights: it confirms ollama serve answers on 11434 (launching it if not) and registers the provider under the tag the endpoint itself reports — the blob's sha256 name would never match anything, since 11434 lists qwen3:14b, not sha256-a8cc…. Two real-machine-only traps fixed on the way: existsSync('<file>/') returns true on Windows, so manifest-vs-directory branching now goes by parseable content instead of trailing-slash semantics; and the quantization label table had been copied from a stale ggml enum generation, labelling ggml_type 15 as Q5_K_S while ollama show reports Q4_K_M. Same-blob dedup is by (dev,ino), keeping whichever entry has a name. Verified on the reference machine: the scan goes from 1 row to 4 — qwen3:14b (8.64G) and qwen3:8b (4.87G) now appear with Q4_K_M badges — in 4ms; npm run verify exits 0 (launcher 57, portable 35, panel-host green including the linux/darwin simulated-platform runs); the new panel-host suite builds a real miniature Ollama store in tmpdir and points OLLAMA_MODELS at it, so results no longer depend on whether the test machine happens to have Ollama installed. From 0.4.6 the panel shows each model its real name instead of its file name, because some download tooling stores every weight under one generic file name, leaving the real name only inside the GGUF header. The scan now reads each file GGUF KV section - not a fixed-size head buffer, because tokenizer.ggml.tokens (hundreds of thousands of strings on a large MoE model) and the chat template push general.file_type past any window; and GGUF string arrays are interleaved (u64 length plus payload per element), so treating them as a length table walks the cursor off the rails and produces astronomically large reads. Reading and skipping are accounted separately: key names and types are always read, string values over 256KB and string arrays are only walked past without copying, non-string arrays over 64 elements are skipped outright, and a cursor past the file size stops the read, because an out-of-range position means the cursor is wrong rather than that a limit should be raised. The displayed name goes Ollama tag, then general.basename plus size_label, then general.name, then the file name, with the full string kept for the hover text since it usually carries internal build numbers; mmproj files are now identified by general.type as well as by name. A short read (truncated file) no longer feeds half a value into the parser. On the reference machine, reading one file header parses 44 keys in about 16ms, and a whole scan takes about 31ms. From 0.4.7 the smoke test no longer fails the first time and passes the second. The old code gave both request attempts one shared AbortController and one shared 30-second timer, so the total budget covered both rounds: a cold load of a large model burned it in round one, the abort left the signal permanently aborted, and the second round (the Ollama-style reasoning_effort: none retry) could not even be sent - which is exactly why the first click returned This operation was aborted and the second succeeded. Now each attempt gets its own controller and timer, a timeout retries once automatically (retryOnTimeout, on by default) because a first-request timeout almost always means the weights are still being read from disk, and non-timeout errors (refused connection, HTTP error) are not retried at all. Timeouts report themselves: the message says the wait exceeded smokeTimeout, whether a retry already happened, and points at the new smokeTimeout setting (seconds, default 60, clamped to 5-600) - port responding is not the same as tokens coming out, so it is a separate knob from launchTimeout. Unrelated: an example naming one specific model was dropped from the panel copy, since not every machine has that model downloaded, and the explanation now describes the mechanism instead. From 0.4.8 adopting a plain .gguf no longer registers an unrelated Ollama model. The port field used to follow the last click rather than the file: selecting an Ollama entry moved it to 11434, selecting a plain .gguf did not move it back, so a 11434 left over from an earlier click stayed. The server accepted it as given, probed the port, found Ollama answering and borrowed it (the skipped-launch log line about the port already answering is this path), and then the model id - derived from the file name - could not be matched against Ollama's /v1/models, where the fallback was to grab the first entry in the list. Embedding models tend to sort first, so the adopt registered a model the user never selected. Now the port follows the file: a plain .gguf landing on another engine's port (ollama 11434, vLLM 8000, LM Studio 1234) is moved to 8090 and the panel says so. Only ports belonging to a known engine are overridden - a hand-typed 9999 aimed at your own llama-server is left alone. The model-id fallback no longer guesses either: it takes the only entry when there is exactly one (a single-model backend may legitimately report a different id than the file name) and otherwise reports an error listing the ids it did see, rather than registering a wrong provider. From 0.4.9 the panel can no longer vanish. It used to be possible for the whole section to disappear from the plugin detail page while the host stayed up, the HTTP bridge kept answering, and no crash log appeared. The cause was a scope error: a helper was declared inside one component and referenced from another, which is valid JavaScript and passes a syntax check, so it only threw at render time - when the user pressed the scan button and React ran the state updater. The host wraps every slot entry in an error boundary whose fallback is an empty anchor, so the throw was contained silently: the panel was not uninstalled or crashed, it was replaced by an empty div. The helpers now live in the module scope both components share. The client half also gained its own test - a minimal React stand-in loads the real panel module, really calls the components, really runs the state updater at render time, and presses the scan button, so this class of error can no longer reach a release.

Matrix-style binary (0/1) digital rain for the Harness interface: columns spawn only while a task is running, fall off screen when it stops, and leave a blank canvas; frosted-glass token layer with adjustable digit and panel opacity.

DSH용 앱 내 브라우저입니다. 지속 프로필을 가진 전용 Chromium을 사용해 Feishu 또는 SSO 로그인이 재시작 후에도 유지되며 CDP로 제어합니다. 에이전트 도구 7개(status, open, eval, read, feishu, screenshot, tabs)를 제공하고 페이지를 오른쪽 사이드바에 실시간 미러링하며 도구 호출 시 패널을 자동으로 엽니다. 가상 스크롤을 훑어 Feishu 문서를 Markdown으로 변환합니다. 로컬 설치된 Chrome, Edge, Chromium 또는 Playwright Chromium 빌드가 필요하며 다운로드하지 않습니다. 특정 바이너리는 DSH_IAB_CHROME으로 지정합니다.

REA의 리버스 엔지니어링 도구 207개(Ghidra / Hopper / IDA 백엔드)를 명령 3개와 읽기 전용 도메인 항목 6개로 묶어 상주 도구 텍스트를 757 KB 대신 약 5 KB로 줄입니다. 엔진은 포함하지 않습니다. rea-agents, Ghidra 12.1.x, JDK 21은 별도로 설치합니다.

composer용 화면 캡처 도구입니다. 화면을 고정하고 영역을 드래그해 선택·주석 처리한 뒤 결과를 메시지 입력란에 넣습니다. Windows 전용입니다. 선택기는 번들 WinForms 바이너리이며 없으면 포함된 C# 소스에서 다시 빌드하고 내장 .NET Framework 4.x를 사용합니다.

Bilibili 영상을 다운로드하고 로컬 whisper.cpp로 전사한 뒤 DSH 자체 LLM으로 구조화된 Markdown 노트를 만드는 Agent 도구입니다. PDF 출력은 선택 사항. Windows 전용이며 의존성이 없으면 data 디렉터리에 자동 다운로드합니다: yt-dlp.exe와 GitHub/gyan.dev ffmpeg.exe 빌드, GitHub whisper.cpp x64 바이너리, hf-mirror의 약 547MB 모델. PDF 내보내기는 로컬 Edge/Chrome을 호출합니다. 도구는 video_to_notes와 note_to_pdf 두 개. 순수 JS, 추가 API 키 없음.

DeepSeek Harness 원격 Workspace: 사이드바에서 SSH 호스트를 선택하고 SFTP로 서버 디렉터리를 Workspace에 마운트해 기본 file tools로 읽기/쓰기/편집합니다. 원격 감시는 polling으로 파일 트리를 새로 고치며, 이미지/바이너리 미리보기는 byte-range 읽기를 사용합니다. `remote_exec`, `remote_search`, `remote_git`가 원격 shell/검색/Git 조회를 제공합니다. 선택적 패치는 에이전트의 bash/pwsh 명령을 서버로 보냅니다. 모든 원격 쓰기/명령은 세션 sandbox 모드와 승인 정책을 따릅니다. 호스트는 플러그인이 암호/키/SSH agent로 관리합니다.

플러그인 저장소의 기존 스크린샷으로 6초 무음 반복 hero reel을 만듭니다. 실제 alpha가 유지되는 VP9 webm, 배경 처리 gif, 커버 프레임, 장면별 유지 시간과 sha256 manifest를 출력합니다. 정수 유지 시간과 전환의 합이 `durationMs`와 정확히 같고 마지막 장면에서 첫 장면으로 이어져 t=0과 t=duration이 같습니다. 런타임 종속성 없이 node:zlib PNG 처리와 SDF 카드 합성을 사용하고 외부 바이너리는 실행 시 찾아지는 ffmpeg뿐입니다. 선택/순서 변경/생성/README markdown 복사 패널과 `storyboard.json` 규약 CLI도 제공합니다.

pen.dev 데스크톱 앱의 MCP 서버를 DSH 세션에 연결해 모델이 Pen에서 열려 있는 `.pen` 문서를 읽고 편집하게 합니다. `get_app_state`, `read_skill`, `execute`, `get_style`, browser 도구를 사용합니다. 일반 설치 경로를 탐색해 서버 바이너리를 찾고, pen.dev 요금제 제한 오류를 짧은 영/중 알림으로 바꾸는 listener와 pen-dev skill도 추가합니다. 비공식이며 pen.dev 바이너리는 포함하지 않습니다.

에이전트가 턴을 실행하는 동안 항상 위에 고정되는 진행 막대를 표시합니다. 비율은 도구 호출 수로 추정하며 턴이 끝나면 100%가 됩니다. DeepSeek 피크/오프피크 요금 시간대를 배지로 표시하고 대량 토큰 작업은 피크 시간에만 확인을 요청합니다. 별도 Electron 프로세스로 실행되며 Electron 바이너리가 없으면 건너뜁니다.

오른쪽 사이드바 문서 미리보기에서 STL 메시를 엽니다. 브라우저에서 바이너리/ASCII를 파싱하고 WebGL 뷰포트와 크기, 부피, 표면적, 수밀성 통계를 표시합니다.

macOS DeepSeek Harness에 doc_ocr_sign_fix 도구를 등록한다. hardened runtime/library validation으로 번들 Python이 onnxruntime/cv2/pypdfium2 native extension을 읽지 못하는 문제를 진단·수정한다. disable-library-validation entitlement를 주입하고 interpreter를 재서명한 뒤 import를 검증한다. 반복 실행 안전, 원본 바이너리를 먼저 백업하며 바깥 앱 bundle은 건드리지 않는다.

macOS arm64 전용. dsh-pet 설치가 필요하다(포함된 바이너리는 소스가 공개되지 않은 사전 빌드). 데스크톱 전환기로 반려동물을 한 번에 표시/숨기며, 앱이 닫혀 있으면 DeepSeek Harness를 시작할 수 있다.

오른쪽 사이드바 문서 미리 보기에서 STL 메시를 엽니다. 브라우저에서 binary/ASCII를 파싱하고 WebGL 뷰포트와 크기·부피·표면적·watertightness 통계를 표시합니다.

ZTE Nde 데스크톱용 DSH Cordis computer-use 공급자입니다. npm install의 postinstall이 computer-use-linux-zte-nde가 빌드한 musl 바이너리(v0.7.7-nde.3)를 패키지 자체 bin/에 가져옵니다. resolveCommand()는 import.meta.url 기준으로 cordis 명령 경로를 해석해 dsh web의 cwd와 무관하게 mount가 작동하게 합니다. ctx.computerUse.register로 공유 computer-use 슬롯을 예약하고 mcp__computer-use-linux-nde-mcp__* 도구(screenshot, list_windows, click, type_text, doctor 등)를 노출합니다. 상류 Nde의 주요 수정은 창 대상으로 KWin caption+coords fallback을 사용하고, SnapshotCache가 (app, pid)별 AT-SPI를 한 번만 깨워 Chrome을 반복 방해하지 않게 하는 것입니다.

모든 요청에 필수 7단계 계획과 단계별 이진 acceptance criterion을 둡니다. worker subagent로 실행하고 독립적인 memoryless verifier subagent가 증거와 함께 PASS/FAIL을 판정합니다.

채팅에서 ployenv 로컬 런타임 버전 관리자를 조작하는 DSH 호스트 플러그인입니다. 런타임 목록/설치/사용, ployenv 바이너리 제공, 세션 셸에 경로 목록 주입을 지원합니다.

리버스 엔지니어링 작업대입니다. 로컬 IDA Pro를 `@deepseek-ai/dsh-mcp-client`를 통해 DeepSeek Harness에 연결하고 `ida-pro-mcp` 도구(decompile, xref, rename, comment)를 `mcp__ida__*`로 노출합니다. `idalib` 헤드리스 모드(IDA GUI 없이 프로세스 내 바이너리 분석) 또는 GUI 모드(포트 13337의 IDA 플러그인 RPC 연결)를 자동 선택합니다. `re_check`, `re_open`, `re_status`도 추가합니다. 번들 MCP 서버는 어떤 MCP 클라이언트에서도 쓸 수 있는 무의존성 stdio 프록시입니다.