플러그인

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.

Session cost meter forked from dsh-cost-meter, retargeted at OpenDesign Cloud: per-session and per-day cost, a 39-model OpenDesign price catalog, OpenDesign Cloud quotas as rolling 5-hour and 7-day used/total bars read from the official funding-check endpoint, Coding Plan quotas, provider balances, peak/off-peak notices, bilingual UI. The quota card needs an OpenDesign Cloud account (the plugin has a built-in login), and it provides service:cost-meter like Han-1413141/dsh-cost-meter and the other dsh-subscribe-meter variants, so install only one cost meter: it replaces dsh-cost-meter rather than running alongside it.

Session cost meter forked from dsh-cost-meter, retargeted at OpenCode Go: per-session and per-day cost, a 90+ model price catalog with automatic matching, OpenCode Go subscription usage as rolling 5-hour / weekly / monthly percentages with progress bars, Coding Plan quotas, provider balances, peak/off-peak pricing notices, bilingual UI. It provides service:cost-meter like Han-1413141/dsh-cost-meter and the other dsh-subscribe-meter variants, so install only one cost meter: it replaces dsh-cost-meter rather than running alongside it.

Session cost meter forked from dsh-cost-meter, retargeted at a self-hosted gateway: per-session and per-day cost, a 90+ model price catalog with automatic matching, gateway quota (one-api / new-api / sub2api, auto-detected) shown as used/total bars for the provider configured in DSH settings, Coding Plan quotas, provider balances, peak/off-peak pricing notices, bilingual UI. It provides service:cost-meter like Han-1413141/dsh-cost-meter and the other dsh-subscribe-meter variants, so install only one cost meter: it replaces dsh-cost-meter rather than running alongside it.

Session cost meter forked from dsh-cost-meter, retargeted at Command Code GOAT: per-session and per-day cost, built-in 90+ model price catalog with automatic matching, GOAT subscription quotas (5-hour / weekly / monthly pool) as used/total bars, Coding Plan quotas, custom provider balances, peak/off-peak pricing notices, bilingual UI. It provides service:cost-meter like Han-1413141/dsh-cost-meter and the other dsh-subscribe-meter variants, so install only one cost meter: it replaces dsh-cost-meter rather than running alongside it.

A Settings page for the memo plugin suite in DSH Web: edits the memo Host's AI route, archived-session summary language, length, transcript budget, template and prompt, and repository URL, adds a GitHub intake section for the claim label, per-run issue cap, autopilot switch and its working directory, preset, time zone and dispatch prompt, with per-field reset and saves that reach the running services without a restart, and carries an update entry that checks the npm registry and can install a newer release through the deployment's plugin manager.

A Workspaces-native Recent sessions section for the DSH left sidebar: favorites, running, waiting-for-you and unread sessions at a glance, with a one-click ☆ on every session row, one-line collapse and in-section search. It takes over the sidebar shell (the shipped sidebar package is disabled) and renders the official workspace/session browser inside it; its recent-sessions panel id conflicts with huangxp12/whybebabo, so do not install them together.

Live step pill in the DSH session header: folds the durable session event stream into the agent's current action (tool name plus its argument summary), the step index inside the running turn, the elapsed turn time and the todo progress, polls it every 1.2 seconds, and renders blue pulse while working, green for a finished turn and red for a failed tool call; hides itself 9 seconds after a turn ends. On Windows it also opens an always-on-top desktop HUD (set DSH_AGENT_STATUS_HUD=0 to disable).

Plugins 페이지 탭에서 DSH 플러그인을 설치·업데이트·제거합니다. 설치는 profile의 내부 pluginManager 서비스를 호출하므로 데스크톱 앱 실행 중에도 CLI 없이 추가할 수 있습니다. 업데이트는 dsh-market 공개 update API v1에 위임하며 해당 섹션에 필수입니다. 시장이 보고한 단계, 비율, 다운로드 바이트를 진행 막대로 표시하고 시장이 restorepoint를 보고할 때만 롤백합니다. bundle 사용 중이라 제거가 거부되면 bundle을 비활성화하고 profile manifest에서 제거하며 남은 파일은 다음 시작 후 정리하도록 예약합니다.

Obsidian 플러그인이 DSH 세션을 제어하는 loopback HTTP/SSE 서비스입니다(목록, 생성, 전송). 최대 10개의 vault 노트를 system prompt에 주입합니다. Obsidian 내부에서 동반 Bridge 플러그인을 실행해야 합니다.

사이드바 패널에서 전체 작업 공간을 이동, 재구성, 병합, 이름 변경, 삭제합니다. 이동하면 각 세션 로그를 새 디렉터리를 헤더에 넣어 다시 게시하고 작업 공간 기록을 수정하며 캐시된 제목을 새로 고칩니다. 카드마다 세션이 실행 중인지 단지 열려 있는지 표시합니다.

Windows Explorer 컨텍스트 메뉴에서 실행 중인 DSH Web 작업공간에 폴더를 추가하고 필요하면 dsh web을 자동 시작합니다. 메뉴를 설치하려면 저장소 클론에서 사용자별 PowerShell 설치 프로그램을 한 번 실행해야 합니다(관리자 권한 불필요).

화면 모서리에 컷아웃 스타일의 whale-girl 메이드 데스크톱 펫을 표시합니다. 클릭하면 DeepSeek 계정 크레딧과 현재 세션 상태를 볼 수 있습니다.

이미 실행 중인 NetEase Cloud Music 데스크톱 클라이언트를 CDP로 제어합니다(시작 시 `--remote-debugging-port=9222` 필요). NetEase 공개 웹 API로 노래를 검색합니다. API 키나 추가 로그인이 필요 없고 회원 전용 기능은 클라이언트에 로그인된 계정을 사용합니다.

실행 전에 모델로 파괴적인 shell 명령을 검토해 사용자에게 확인하거나 차단합니다. 실행 중인 명령이 너무 많은 파일을 삭제하면 중단합니다.

DSH Web UI에 Cursor 스타일 모드 전환을 추가합니다. plan mode에는 위임 gate, multitask mode에는 tool-call quota가 있으며 create_plan/request_more_edits/list_running_agent 도구를 제공합니다. Build 버튼이 있는 plan card를 추가하고 하위 에이전트 드롭다운은 실행 중 항목만 표시합니다.

Windows 전용: 세션 헤더에 세션별 자동 종료 스위치를 둡니다. 활성화된 모든 세션에서 턴·작업·서브에이전트가 실행 중이지 않고(기본값은 다른 모든 활성 세션도 유휴 상태여야 함), 2분 카운트다운 후 PC가 종료됩니다. 카운트다운을 누르면 취소되며 스위치가 없는 세션은 카운트다운을 시작하지 않습니다.

DeepSeek Harness의 하위 에이전트와 Agent Teams 팀원을 관리합니다. 실행 중인 인스턴스를 멈춘 다음 카탈로그와 roster 기록을 순번 보존 로그 tombstone으로 은퇴시키고 세션 디렉터리를 복구 가능한 휴지통으로 옮겨 실제 제거합니다. model tools, 대화 헤더 패널, 오프라인 CLI를 제공합니다.

5가지 DSH 유틸리티를 한 플러그인에 제공합니다. QQ로 기존 세션을 제어( NapCat 같은 OneBot 11 구현을 함께 실행해야 함), context guard를 둔 다른 세션 질문, 실행 중인 턴 일시 중지, Windows에서 DSH 재시작 및 선택적 자동 재개(install-restart.ps1 예약 작업 이용), 혼잡 시간대 토큰 절약 규칙 주입 기능입니다.

DSH를 로컬 BiliMemory backend에 연결해 Bilibili 검색/컬렉션/상태/로그인 작업, A-Memorix 공유 메모리 상태 확인, QQ 연동 상태/허용 목록/watcher 재시작을 지원합니다. Windows 원클릭 installer는 없으면 MaiBot A-Memorix를 설치/복구한 뒤 local-bge-small-zh, embedding, 공유 메모리를 설정합니다. 메모리만 복구하는 `install-memory.cmd`도 제공합니다. 요구 사항: Windows, DSH Desktop 0.2.1-rc.2+, Node.js 22.19+, Python 3.12, Playwright/Edge, FFmpeg, `127.0.0.1:8765` BiliMemory API, `127.0.0.1:11435` 로컬 embedding 서비스, Bilibili 로그인, QQ용 MaiBot/NapCat. Voice, RVC, cover 기능은 제외됩니다.

Windows/Linux의 로컬 NTQQ(QQ 9.9.x) DB를 읽어 그룹/개인 채팅을 txt/csv/json/md/html/sqlite로 내보냅니다. 도구 16개: qqnt_status, qqnt_doctor, qqnt_extract_key, qqnt_index, qqnt_list_groups, qqnt_list_chats, qqnt_search, qqnt_stats, qqnt_export_group, qqnt_export_chat, qqnt_export_batch(모호한 이름은 추측하지 않고 거부), qqnt_jobs, qqnt_media, qqnt_storage(디스크 사용량/허용 목록 기반 정리), qqnt_ask, qqnt_recall(인용 검색). 로컬 UI 127.0.0.1:46728은 증분 SQLite 색인, 6개 보기, 작업 패널, 좁은 화면의 단일 창, 발신자 비식별 처리, 미디어 묶음, 날짜별 분할, 내보내기 프리셋, AES-256 암호 ZIP을 제공합니다. OCR·음성·임베딩은 필요할 때 플러그인 전용 virtualenv에 설치됩니다. python3와 sqlcipher3가 필요하며 설정에서 virtualenv를 만들 수 있습니다. DB 키는 실행 중이고 로그인된 QQ 클라이언트에서 읽습니다.

AutoCAD, ZWCAD, GstarCAD, BricsCAD를 MCP로 DeepSeek Harness에 연결합니다. CAD server를 포함하며 첫 실행 시 전용 Python 환경을 만듭니다. Windows, CAD 앱 중 하나의 설치, PATH에 있는 Python 3.10 이상이 필요합니다. CAD는 DSH와 같은 Windows 사용자 및 권한 수준에서 실행해야 합니다.

DeepSeek Harness 세션을 보관하기 전에 확인하고 사용자가 고른 디렉터리에 백업합니다. 세션마다 폴더 하나를 만들고 session log, 첨부 파일, metadata, manifest, 복원 안내와 Markdown ledger를 저장합니다. 오른쪽 사이드바에 보관 세션 브라우저와 더블클릭 재시작 버튼을 추가합니다(Windows 전용, 세션 실행 중 비활성화).