플러그인

A sticky-notes wall pinned to the sidebar: one click opens a full-page wall, notes save as you type, and writing #tags in the text turns them into one-click filter chips. Six paper colors, pinning, search, JSON export and import.

Hide the duplicate DeepSeek account / official API entries, sidebar items and the Feedback / Sign out items; on DSH Desktop, add Reload / Restart to the application menu with a one-click patch to the app shell main.js (needs an app restart, and a DSH update reverts the patch); pin a reasoning level and auto-fill new providers and models.

Purge DSH sessions for good and manage them from one console: per-session deletion, a filterable bulk console, and housekeeping sweeps.

Reference links panel for DeepSeek Harness: collects every URL of a conversation into the right sidebar with pins, turn counts, metadata and favicon caching. DSH 右栏「参考链接」面板。

Drive a real Android phone from DSH: screenshots, taps, swipes, typing and app launching. Requires Python 3.10+ with aiohttp and mcp on the computer side, plus the AutoX.js script running on the phone with screen-capture and accessibility permissions; the phone long-polls your machine, so no public port and no relay server.

Runs project check commands itself and records the real exit codes, pins the result to a git or file workspace fingerprint, expires the evidence when the tree changes, and reminds the agent to verify before claiming completion.

Task alerts for DSH: a system toast when a turn finishes, or when an approval, question or plan review is waiting; a chime rendered on the Windows communications endpoint so the system ducks other media by its own setting and restores it afterwards, without anyone's volume being written; and an optional copy mailed to your phone over SMTP (465 implicit TLS, or STARTTLS where credentials go out only after the upgrade). No runtime dependency: the sound helper is a PowerShell script, the mail client is plain Node.

Domain-routed proxy for DSH fetch traffic: hosts on a force-proxy list always go through the proxy, every other host tries direct first and fails over to the proxy within a second, the first channel to answer is pinned for the session, connections are reused for 120 seconds, and each attempt is logged. Can also route pnpm/git subprocesses used by plugin installs.

Adds a delete row to the menu of an archived conversation only, and after a confirmation removes the session log directory, its projection cache, the archive and pin marks and the workspace booking together.

Archives sessions that have gone unused for longer than a set number of days (5 by default), skipping archived, pinned, and protected ones — a session context-menu item marks a single session as protected, a small shield shows it on the row, and a General settings field changes the threshold.

Adds a confirmed Delete session action to the sidebar menu of archived Sessions; the Host permanently removes their local JSONL logs (including hidden subagent descendants), workspace accounting and archive/pin entries, then refreshes open Web clients.

A right-sidebar tab listing the files DSH Sessions produced, newest first, with the Session in view expanded on top: a click copies a row's full path, a double click previews it in the sidebar, and a file can be pinned in one Session or in all of them.

Caps every image sent to a model at a per-side pixel limit (default 2000 px) by wrapping attachments.readImageRequest, which fixes Anthropic many-image request errors without touching stored images.

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.

Adds a Speed setting (Standard, Fast, Ultrafast) for OpenAI and Codex models on a patched dsh build: the pinned dsh 0.2.1-alpha.2 web client needs the Model / Effort child-slot patch, the host pi-ai must be 0.87.1 or 1.1.0 (other versions refuse to load), and the terminal controls need a patched dsh-TUI. Then it appears in the Model / Effort menu; in dst /speed opens a picker or sets the speed directly, and the current speed shows in the bottom bar. The choice is kept per session, provider and model and is sent as the request service tier.

Provider-first model picker for the composer. Every provider is listed in the left rail and its models in the right pane. Providers, models within a provider, and pins reorder by drag or Alt+Arrow; providers take a local display alias; one provider can be the panel default; a model's reasoning levels expand inline; search matches model names, provider names and aliases. Switching goes through the shipped model directory service, so the /model popup and this panel always agree.

Replaces the DSH sidebar list with a workspace rail: pinned chats and folders, projects that are folders with their own AGENTS.md and files, a virtualized Recents list ordered by the last finished turn, a project_search tool, unread markers, a collapsed icon column with an unread count and a live browser tab title.

Keeps a DSH agent turn open when the model ends it by narrating work it has not done: on agent/turn-stopping the guard replays the turn log and classifies the trailing message with regexes — in the default mode wait / ask / done / gray end the turn with no model call, and only a promise of pending work reaches the judge — then has a judge chain of four dialect kinds (System One three-question noul, OpenAI-compatible, OpenAI Decisions, or the session LLM) rule on whether the turn is really over, and steers the same turn with agent.steer when it is not.

DSH 웹 클라이언트 왼쪽 사이드바에 Skills 패널을 추가합니다. 현재 DSH 인스턴스의 skills를 나열하고 즐겨찾기를 고정하며 /skill-name을 작성창에 삽입합니다.

공유 기본 모델을 변경하고 고정합니다. 새 대화, 재시작, 채팅 안에서 모델을 바꿔도 기본 모델이 다시 덮어써지지 않습니다.

DSH Web UI용 세로형 휴대전화 레이아웃입니다. 왼쪽 가장자리에서 밀면 사이드바가 화면 밖에서 열리는 서랍으로 나타납니다. 작성창 아래에는 56px 작업 표시줄이 고정되고 설정 패널, 모델 및 권한 선택기가 터치에 맞게 재배치됩니다. 기본형, 통합형, 플로팅 레이아웃 프리셋은 html\[data-dshm-mode="mobile"]에만 적용되어 데스크톱, 태블릿, 가로 화면은 기본 레이아웃을 유지합니다.

DSH의 키보드 기반 빠른 전환입니다. 색상 공급자 태그·접미사 태그·고정 기능이 있는 모델 팔레트, 질문별 프롬프트 미리보기와 보관 세션 복원이 가능한 2열 세션 팔레트, 최근순 세션 순환, 다음 답변 완료 항목으로 이동, 작성창 포커스, 설정 가능한 권한 순환을 제공합니다. 사이드바 제어 패널에서 관리합니다.

DeepSeek Harness 세션의 마지막 턴을 그 자리에서 다시 작성합니다. 취소되거나 실패한 프롬프트를 수정해 재전송하고, 이미 작성된 파일은 유지합니다.

DeepSeek Harness용 지속형 pwsh 세션 도구 `pwsh_run`입니다. 이스케이프 없이 PowerShell 스크립트를 실행하고 32K 명령줄 제한이 없으며 호출 사이에 상태를 유지합니다.